This is about the annual report of Baillie Gifford US Growth Trust, a fund that buys American growth stocks. Last year it bounced back strongly, but over the past five years it badly lagged the US stock market. The fund also has a third of its money in private companies (not listed on stock exchanges), which makes its share price trade at a discount to its underlying assets. For ordinary investors: expect high volatility and no dividends, and be cautious about long-term performance. It's worth reading because management openly admits mistakes and introduces a new rule to re-check any stock that triples, showing how a big investor learns from pain.
The fund has significantly underperformed over five years, yet has matched the index since inception; management is rebuilding trust with long-termism as a shield and process improvements as a spear, but the share of private assets and the near-10% discount remain core risks [Cautious]
This section covers the fund's "Introduction" and "Investment Purpose and Principles" statements. Its core purpose is to announce annual results for the year ended May 31, 2025, and reaffirm the long-term growth investing philosophy. The overall tone is the fund speaking for itself, with a clear position-holder perspective and marketing component.
During the reporting period, the fund significantly outperformed its benchmark: total returns on share price and NAV were 24.5% and 22.1%, respectively, versus only 7.2% for the comparator benchmark (S&P 500 total return in GBP).
| Metric | Total return for year ended May 31, 2025 |
|---|---|
| Share price total return | 24.5% |
| NAV total return (borrowings at fair value) | 22.1% |
| Comparator: S&P 500 total return (GBP) | 7.2% |
The original chart shows that during the reporting period the share price traded at a discount to NAV, fluctuating between roughly -3% and -15%, although specific month-end figures are not disclosed in the text. In addition, the report notes that as of May 31, 2025, NAV was identical whether valued on a fair-value or book-value basis.
The author reaffirms his long-term framework here, and the core views can be summarized in six principles:
1. The index is not the construction starting point: The index allocates capital by market capitalization, whereas the author believes capital should be allocated on the basis of marginal returns and growth capacity; large companies are not immune to disruption, and the portfolio's active share is expected to be high.
2. Success is measured over cycles of five years or more: The fund's turnover has consistently been low, consistent with a long-term time horizon.
3. Essentially no preference between listed and private status: Through diligent analysis, capital is deployed to where risk-adjusted returns are highest, which provides the intellectual foundation for investing in private companies.
4. Do not sacrifice long-term returns for short-term volatility: The author believes many managers accept smaller but predictable risks because of loss aversion, which damages long-term returns; therefore, its short-term performance may be "lumpy" (uneven).
5. Maximize shareholders' long-term wealth, rather than short-term performance.
6. Maintain low costs: The author stresses that expense compounding erodes returns, and believes academic research shows that long-term equity returns are dominated by a small number of exceptional growth companies, and most stocks are not worth holding.
The author crystallizes his view of risk in one line: "the costliest of mistakes is excessive risk aversion." At the same time, the author acknowledges, "we will make mistakes," attributing this to the inherent asymmetry of equity markets — when right, one earns far more than one loses when wrong.
It should be flagged that this chapter is the fund's own statement; all the principles serve to rationalize its long-term growth strategy, and the reader should treat it as a position-holder's perspective rather than independent analysis. Apart from the table-of-contents structure, this chapter does not address specific stocks, position changes, or capital-structure information.
The annual report skillfully splits performance data into two coordinate systems, presenting two very different narratives.
Five-year dimension: Share price return 26.7%, NAV return 45.4%, versus 92.1% for the index over the same period. This is the narrative that requires acknowledging mistakes; both the Chairman and the managers used words such as "disappointing" and "below expectations."
Since-inception dimension (March 23, 2018 to date): NAV return 170.0%, almost exactly matching the S&P 500 index's 170.1%. This is the narrative to be proud of — "We are satisfied to have matched a period of strong growth from the index since inception."
The problem is that the gap between these two data sets is far from as downplayed as it appears on the surface.
| Period | Annualized NAV return (approx.) | Annualized S&P 500 (GBP, approx.) | Annualized excess |
|---|---|---|---|
| Since inception (approx. 7.2 years) | ~15.6% | ~15.6% | ~0 |
| Recent 5 years | ~7.8% | ~13.9% | ~-6.1% |
| From inception to 5 years ago (approx. 2.2 years, estimated) | ~33% | ~17% | ~+16% |
Two straightforward inferences:
1. The COVID base effect is real, but its explanatory power is limited. During the roughly two years from inception to mid-2020, the fund significantly outperformed the index, pushing the base to an extremely high level. In the subsequent five years, even if stock selection had only slightly underperformed, the cumulative excess would have been magnified into a significant lag. The manager attributes this to the "base effect," which is mathematically valid — but this only explains "why the five-year numbers look so bad," not "why the fund genuinely underperformed over those five years."
2. More noteworthy is the divergence between share price and NAV. On the five-year dimension, the share price return (26.7%) is nearly 19 percentage points lower than the NAV return (45.4%). Working backward from the period-end discount of 9.4%, the fund must have traded at a premium of roughly 4% five years ago. If this estimate holds, it means the fund experienced a valuation contraction from premium to discount over the past five years — the market's pricing of the fund's future growth potential underwent a change in attitude. Although the Chairman emphasizes that the discount narrowed from 11.2% to 9.4% this year, this only repairs a small portion of the valuation damage accumulated over many years.
The allocation to private (unlisted) companies rose from 34.1% to 34.9%, with 27 companies held and three new names added (Cosm, Rippling, Runway AI). This data point is directly linked to the discount issue, and the annual report obscures its complexity:
Under this structure, the 5.4% buyback (£35.5 million) looks more like a defense against the discount than a cure. When more than one-third of the portfolio consists of illiquid assets, buybacks can alleviate supply-demand imbalances but cannot eliminate the underlying valuation-uncertainty premium. Notably, with the discount still near 10%, the buyback is equivalent to purchasing £1 of NAV for £0.906, which is accretive for continuing shareholders — a point the annual report does not quantify.
Net income per share was -2.54p (prior year: -2.07p), with the deficit widening by around 23%. A zero dividend, for a UK-listed investment trust, openly challenges the core expectations of a portion of the shareholder base.
There is a structural fact here worth digging into: management fees are charged on total assets, while a large portion of the portfolio's private company investments themselves generate almost no cash flow. This means that as the portfolio grows and fees rise in tandem, the income side has very little room for growth. The widening deficit is not accidental; it is an inevitable consequence of this strategy model.
The Chairman said that if future income increases, the company will "seek to distribute a dividend at the minimum viable level to maintain investment trust status." That statement pragmatically acknowledges reality: dividends are not the objective; maintaining tax status is the only driver of any distribution. For investors who rely on income or hold the fund outside a pension account, this is a product feature that should be clearly understood.
The manager disclosed a specific change in investment process: once any listed stock's return exceeds 2.5x, regardless of time horizon, it will automatically trigger a review of that position.
This mechanism deserves interpretation from a strategic-logic perspective. The annualized implication of 2.5x depends on time: reaching 2.5x in 10 years is roughly 9.6% annualized; reaching 2.5x in 2 years is roughly 58% annualized. The real intention of the automatic re-review is not to prevent positions from "rising too much," but to prevent individual stocks that experience extreme gains in a very short period from obtaining portfolio weight disproportionate to their fundamentals because of momentum in market sentiment. This is an ex-ante systematic constraint — a rule-based intervention precisely at the moment when human judgment is most susceptible to temptation (large unrealized gains).
There is a noteworthy tension between this mechanism and the fund's "high-conviction, high-growth" positioning: it implies the fund acknowledges that its most successful stock picks can also become its biggest source of risk. The manager said, "the volatility that we delivered during and after the pandemic exceeded our expectations" — translated, this means that the fund previously allowed extreme gains in a single position or in related positions to push portfolio volatility beyond an acceptable range. The 2.5x re-review mechanism is the structural answer to that problem.
On December 18, 2024, the company received a formal requisition from Saba Capital Management to call a general meeting; the meeting was held on February 3, 2025, and the Board claimed it received "overwhelming shareholder support."
The Chairman's statement is, in the governance context of the annual report, unusually vague: there are no precise figures for turnout or the proportion of votes for and against. For a trust that encourages shareholders to vote with an "Every vote counts" stance, avoiding specific voting results is notable. UK-listed companies customarily disclose voting results via RNS after significant general meetings, and annual reports typically repeat the specific proportions when citing them. "Overwhelming support" is an evaluative phrase; it is legally accurate, but it falls short in completeness of disclosure.
Another detail that is glossed over: after the activist shareholder intervention, Board member Palmer promptly left "to pursue other opportunities." The annual report provides no additional explanation of the relationship between her departure and the Saba episode, nor does it mention whether she was challenged by shareholders in the February vote. Departures following activist involvement are usually accompanied by an explicit statement that it was "not due to Board disagreements" — the silence here is itself information.
The ESG section consists of just one paragraph. Its core argument is that "sustainability and long-term investing are inseparable," and it notes that the companies invested in are mostly "asset-light, low-carbon-footprint" businesses.
This handling is consistent with Baillie Gifford's public stance: it refuses to turn ESG into a scorecard, instead embedding it within the long-term growth logic. Against the backdrop of the UK FCA's SDR rules (Sustainability Disclosure Requirements) becoming increasingly stringent and investor demands for sustainability disclosure rising, a one-paragraph ESG discussion looks thin — but this may be precisely the fund's deliberate positioning: it makes no sustainability-label commitments, in order to avoid taking on additional compliance burdens under the regulatory framework.
The narrative structure of the full text can be summarized as: acknowledge short-term underperformance, emphasize long-term repair, and trade process improvements for investor trust.
This is a fairly complete investor-relations narrative. But for investors doing deeper analysis, the key figures — the Saba vote results, the specific magnitude of the premium five years ago, the valuation-discount policy for private company investments, and the context of Palmer's departure — remain gaps that need further questioning.
The previous chapter pointed out that investing cannot stop at visible architecture (such as business models and market share) but must go deeper into the cultural foundation that supports these structures. This section further unpacks the dichotomy of "root culture" and "generative culture" presented in the report, and adds a frequently overlooked middle layer: management's "scaffolding" role. The report notes that the management team cannot directly control how root culture manifests, but can use indirect means — incentive mechanisms, organizational structure, decision-making processes — to allow core beliefs to permeate the organization. This view can be tested quantitatively: for example, when CoStar founder Florance acquired Apartments.com in 2014, the platform had revenue of only $85 million; today its revenue scale has surpassed CoStar's original data business and become the group's largest revenue unit. This leap stems both from adherence to the underlying logic of real estate data (root culture) and, even more, from the "generative culture" operation of quickly merging the original data team with the rental-platform team after the acquisition — without the latter, daily execution details such as data integration, interface reconstruction, and sales-incentive reshaping could not have landed.
| Company | Root culture ("how things should be") | Generative culture ("how things are now") | Scaffolding measures |
|---|---|---|---|
| CoStar | Break down data silos, build interconnected real-estate information | Rapidly integrate data and sales teams after acquisition | Upgrade Apartments.com from a pure rental listings platform to an integrated transactions-and-data platform |
| Shopify | Merchant success is the first priority | Gradually add payments, capital, and fulfillment capabilities | Decisively exit logistics in 2023, pivoting resources into AI R&D |
| Stripe | Expand the GDP of the internet | Integrate stablecoin rails into the payments stack | Acquire Bridge for its technical team and compliance licenses, rather than building in-house |
| Duolingo | Make language learning accessible to everyone | Use AI character Lily for conversation practice | Embed generative AI into subscription products, directly lowering marginal tutoring costs |
| SpaceX | Make humanity a multi-planetary species | Reusable rockets + rapid iterative testing | Adopt a "fly-test-fail-fix" loop in the Starship program, rather than theoretical rehearsal |
These cases show that excellent "builder" CEOs not only plan the long-term vision, but also know how to design adjustable scaffolding at the "generative culture" level. For example, Shopify's Tobi Lütke exited the logistics business in 2023. On the surface it was a strategic retreat, but the deeper reason was that the logistics business conflicted with the "merchant-first" root culture — the heavy assets and low flexibility of in-house logistics actually weakened the merchant experience. His reallocation of resources toward the AI assistant Sidekick is a typical move of strengthening root culture by adjusting "generative culture."
The report emphasizes that Trump's tariff policy is short-term noise, while the real drivers are structural trends such as falling battery costs, falling AI inference costs, and biomedical breakthroughs. A quantitative perspective can be added: according to BloombergNEF data, the average lithium-ion battery pack price has fallen from $1,100/kWh in 2010 to $115/kWh in 2024, a cumulative decline of roughly 90%. On AI inference costs, according to the Stanford AI Index, from GPT-3.5 at the end of 2022 to GPT-4-class models by mid-2024, the cost per inference fell by more than two orders of magnitude. These trends are unaffected by political cycles; they are precisely the "price-insensitive" long-term forces.
In contrast, short-term policy shocks often cause excessive stock-price reactions. The report cites Tesla's market capitalization rising by more than $500 billion in the month after Trump's election as a vivid example. One can compute: before November 2024, Tesla's market cap was around $700 billion, and within a month it rose above $1.2 trillion, an increase of more than 70%. But did this reflect fundamental improvement? During the same period, Tesla's vehicle sales grew modestly; the main driver was expectations of regulatory relaxation (such as autonomous-driving approval). When expectations run ahead, the risk-reward ratio deteriorates — and this is the cultural decision behind the investment team's trimming of the position: they are not predicting Trump's "next tweet" but assessing Tesla's true position in long-term trends (such as full self-driving and robotics). This echoes the report's earlier logic that "the management team cannot control how root culture manifests": investors likewise cannot control market sentiment, but they can control their own analytical framework.
The report compares CEOs to builders who are willing to "build extension structures that are occasionally unpopular" and also willing to "dismantle parts of the building when necessary." This view is worth extending: the essence of the builder mindset is not blind expansion, but indifference to sunk costs and a pursuit of option value.
Take SpaceX. The Falcon 9's reusable technology upended launch economics, but Starship has required enormous R&D investment and has suffered multiple test-flight failures. By the end of 2024, Starship had completed eight test flights, and each flight collected new data. This culture of "rapid iteration + tolerance of failure" is precisely the combination of root culture ("make humanity a multi-planetary species") and generative culture (the "fly-test-fail-fix" loop). In contrast, traditional aerospace companies such as Boeing adopted a "pursue first-time success" strategy for the Starliner program, and instead suffered repeated delays because of excessive caution. This contrast shows that the "build-and-dismantle capability" of the builder mindset also includes dismantling R&D processes that were once effective but have become outdated.
Another underestimated example is DoorDash. Its core capability is "real-time optimization of delivery logistics," but the company is extending this capability from restaurant delivery to high-frequency non-restaurant categories such as convenience stores, groceries, and pharmacy. This expansion seems aggressive, but it is actually deepening the moat of "driver-density economics": the richer the variety of categories within a delivery area, the lower the average delivery cost per mile. DoorDash's root culture — "empowering local economies" — is made concrete here: it is no longer simply a delivery company, but the digital nervous system of local commerce.
The report discloses an annual turnover rate of only 9.1%, which is low among active management funds. Low turnover usually implies high conviction, but it can also conceal excessive concentration. However, against the dual backdrop of "generative AI" and the "Trump shock" over the same period, 9.1% turnover is not conservatism but selective response. One can calculate: 8 new buys and 4 full sells add up to 12 trades, corresponding to roughly 9% turnover, meaning each trade involved, on average, no more than 1.5% of the portfolio. This style of small, multi-step adjustments is closer to "rebalancing" than "reconstruction," reflecting a management culture of patience amid change.
Particularly noteworthy is the reduction in Tesla. The report says "the size of our holding no longer reflects our conviction." That sentence looks vague, but it actually sets out a behavioral rule: conviction must match position size, rather than simply holding forever. This differs from Warren Buffett's maxim, "If you aren't willing to own a stock for ten years, do not think about owning it for ten minutes." It is closer to "long-term horizon, but adjust when short-term valuation becomes extreme." The cultural basis underneath is: trust the company's fundamentals, but do not be blindly loyal to market sentiment.
Taleb introduced the concept of "antifragility," meaning a system not only withstands shocks but benefits from them. Most of the builder companies in the report display this trait: CoStar increased reinvestment while competitors struggled, Shopify pivoted to AI after its logistics failure, and SpaceX accelerated iteration after test-flight failures. All of these are expressions of "cultural antifragility." Conversely, many so-called "solid" companies collapse when the environment shifts suddenly because their culture has become rigid.
Therefore, for investors, understanding a company's culture matters more than predicting next quarter's earnings. Culture is the foundation; structure is the superstructure — when the foundation is deep enough, builders can continue to erect buildings on uncertain ground. This also explains why the author emphasizes "patiently navigating volatility" and "conviction-based investing": genuine cultural advantages take time to translate into financial results. In a highly volatile market in the Trump era, this patience is scarce, and that is precisely the source of returns.
The position adjustments in NVIDIA, Tesla, and The Trade Desk were not based on deteriorating fundamentals, but on a recalibration of "price vs. value." In the case of NVIDIA, management explicitly stated that "the conviction in AI's disruptive potential has, if anything, grown stronger," yet the stock price had already priced in too much growth expectations — that is, the growth in market capitalization outpaced the speed of earnings delivery. This reveals a key principle: even a great company can be a bad investment in its stock. The Tesla reduction was compounded by the fact that its multiple businesses (autos, energy storage, Optimus) are still early-stage and hard-pressed to support the current valuation multiple. The Trade Desk was removed from the top ten purely on valuation grounds after two meetings with founder Jeff Green. This approach — "willing to talk with excellent management, but ultimately deferring to the math" — embodies rational restraint in active management.
A notable contrast: reducing a position is not the same as clearing it. NVIDIA remains in the top ten, and Tesla also retains a position. The message this sends is that long-term investors should not exit completely at the top of a bubble, but instead manage risk by adjusting weights while preserving the capacity to add on the downside.
| Company | Adjustment | Core reason | Conviction status |
|---|---|---|---|
| NVIDIA | Significantly reduced | Share price has already priced in the opportunity | AI conviction has, if anything, increased |
| Tesla | Notably reduced | Businesses early-stage, valuation too high | Long-term positive, short-term cautious |
| The Trade Desk | Reduced holding | Determined overvalued after two meetings | Respect the company, reject the price |
Cloudflare, DoorDash, and Duolingo entered the top ten through two very different paths:
This reflects two driving forces behind portfolio adjustments: both "let profits run" with the trend, and "add after cognitive upgrade" against the trend. In contrast, Moderna and Brex fell out of the top ten because short-term fundamental pressure (shrinking vaccine market, cost restructuring) led to slowing growth. This is a reminder that the top-ten list is not a permanent seat, but a mirror of a company's competitive position in the current industry cycle.
DraftKings, Lineage, SharkNinja, Globant, and Ensign Group operate in gaming, cold-chain warehousing, home appliances, IT services, and senior care, respectively. They look unrelated at first glance, but a closer look reveals a deeper logic: they are all consolidating in markets with extremely low industry concentration, leveraging technology, scale, or management capability.
These companies are not "glamorous names riding the trend"; they are "bulldozers on muddy tracks." Their common features are rigid industry demand, proven management teams, and moats built gradually through economies of scale.
The fund explicitly states that "there is no single best way to capture AI potential," and instead builds a portfolio across the AI stack:
| Layer | Representative holdings | Role |
|---|---|---|
| Cloud infrastructure and chips | Amazon (AWS), NVIDIA | Provide compute and foundational services |
| Edge inference | Cloudflare | Push AI compute to the user side, reducing latency |
| Data infrastructure | Databricks, Snowflake | Data management and preprocessing, a key prerequisite for enterprise AI implementation |
| Application layer | Runway AI | Use existing infrastructure to develop generative AI tools, empowering creative workflows |
The beauty of this layered allocation is that it does not assume which layer will win, but ensures that the winner of each layer appears in the portfolio. Even if the application layer erupts while infrastructure-layer value is compressed, or vice versa, the portfolio can benefit from structural growth. At the same time, this avoids the risk of any single stock being overvalued — for example, even if NVIDIA pulls back, Cloudflare or Databricks may rise on AI inference demand, creating a hedging effect.
No excellent investor can guarantee that every investment will be right. The right approach is to identify mistakes quickly and reallocate capital:
These sells do not follow a single "pattern," which shows that investment decisions are judgments made in specific contexts, not mechanical rules. The key is to maintain the efficiency of the capital cycle — the proceeds from sales are used to buy new opportunities, not left idle.
In response to the challenge "why not simply buy the S&P 500 index," the report offers a powerful rebuttal:
The key to this argument is that it identifies the hidden assumption of passive investing — that market prices already reflect all information. But truly high-growth opportunities tend to emerge before the market's perception shifts, and this requires active research, patient holding, and long-term structural support.
This passage further reveals the operationalized path of long-termism in Baillie Gifford's investment management — it is not an abstract slogan, but is implemented through three dimensions: governance expectations, voting behavior, and direct engagement. Several implied judgments here deserve to be singled out:
The text explicitly states that "helping management teams resist short-term demands from shareholders often protects returns"—a form of shareholder activism with a clear stance. Unlike traditional passive index funds, Baillie Gifford does not position itself as a mediator of all shareholders' interests, but rather actively sides with management against short-termism pressure. Yet this does not imply unconditional support—the governance paragraph that follows immediately pairs "trust" with "accountability," forming an asymmetric mechanism of trust first, verify later, hold accountable upon breach. This attitude bears some resemblance to Berkshire Hathaway's approach, but is more rules-based: long-term support is premised on management consistently demonstrating "deliberate" prudence and "transparency."
Notably, the text explicitly opposes "destructive financial engineering," which in the contemporary context typically points to excessive share buybacks, aggressive leverage, or growth disguised through spin-offs and acquisitions. Baillie Gifford prefers to allocate capital toward areas that generate genuine intrinsic value—R&D, capacity expansion, talent investment—a stance highly consistent with the public portfolio characteristics of its heavy tech-growth holdings.
Management emphasizes that "no single governance model produces the best long-term outcomes" and explicitly "push back against one-dimensional global governance principles." This is an implicit critique of the standardized governance metrics commonly found in ESG rating systems (such as board independence, term limits, one-share-one-vote, etc.). Baillie Gifford prefers:
This stance is complementary to its decentralized, autonomous investment team culture: since there is no "one-size-fits-all" approach to corporate governance, it is only natural for investment teams to apply different voting standards across companies—and even to cast different votes on the same proposal. The text explicitly records that "investment teams sometimes choose different votes on the same resolution," which is uncommon across the asset management industry but consistent with its long-term, bottom-up investment philosophy.
| Voting Type | Meetings / Share | Interpretation |
|---|---|---|
| Votes aligned with management | 82 meetings | An overwhelming majority, reflecting long-term collaborative relationships |
| At least one vote against / abstention / reservation | 12 meetings | Roughly 13% of meetings featured some dissent, showing genuine independent judgment |
| No vote cast | 3 meetings | Potentially involving conflicts of interest or special circumstances |
| For-vote percentage | 92.0% | Significantly above the industry average (typically 80-90%) |
| Against-vote percentage | 6.0% | Low, but sufficient to draw management's attention |
| Abstentions / reservations | 0.9% | Minimal, indicating a habit of taking a clear stance rather than passively avoiding |
| Not voted | 1.1% | Linked to the meetings with no vote cast |
Compared with the voting data of major U.S. asset management giants (such as BlackRock and Vanguard), Baillie Gifford's opposition rate is slightly lower, but its proactive engagement strategy—communicating privately first, then escalating through voting—means its opposition votes tend to be more consequential than those of index funds, because they typically occur after dialogue has failed, rather than as a mechanical mapping of policy. The 12 meetings with "at least one vote against" correspond to the phrase "voting as an escalation mechanism" in the text, indicating that voting is only the final link in its continuous engagement chain, rather than an isolated event.
The three cases (Aurora, Netflix, SharkNinja) are in vastly different industries, but they embody a common investment-research logic — validating the key assumptions behind a long-term growth narrative, rather than assessing quarterly earnings.
| Company | Focus of Attention | Core Investment Hypothesis | Validation Outcome |
|---|---|---|---|
| Aurora | Scaling challenges after commercializing autonomous trucks | Shift from "proving viability" to "executing volume ramp-up" | Commercial launch, deepened customer partnerships; entering a new execution phase |
| Netflix | Monetization path and timeline for the advertising business | Whether ad revenue can achieve per-user revenue parity | In-house ad tech stack; ad revenue per user expected to catch up within three years |
| SharkNinja | Supply chain diversification and DTC expansion under tariff risk | Whether global capabilities are sufficient to offset geopolitical losses | Approximately 40% of business from outside the U.S.; manufacturing distributed across multiple countries |
All three cases took place in an environment of elevated macroeconomic uncertainty (AI's impact on advertising, trade barriers, autonomous-driving regulation), but Baillie Gifford is not concerned with whether next quarter's profits are impaired; rather, it asks whether management is making key decisions for "sustainable long-term value." For example, with Netflix, it did not push on how fast ad revenue was growing, but instead delved into the strategic significance of building the ad tech stack in-house; with SharkNinja, it tested whether the combination of "multi-country manufacturing + rapid innovation" can withstand tariffs — this is structural assessment, not event-driven analysis.
In addition, it is worth noting that the counterparts in these three cases were mostly CEOs/CFOs, and the discussions leaned toward execution details at the operational level (such as Aurora's resource allocation, Netflix's technology investment, and SharkNinja's factory footprint), which is a far cry from pure financial analysis. Engagement activities here play more of a "deep due diligence" role, helping the investment team confirm whether the case for holding a position still holds.
The report directly links long-term success to the "social licence to operate" and emphasizes complying with "the spirit, not merely the letter, of laws and regulations." This implies that Baillie Gifford, when evaluating portfolio companies, pays attention to the quality of a company's relationships with its communities, regulators, and the public. In a cycle of retreating globalization and tightening regulation, losing social licence can raise real operating costs (e.g., antitrust breakups, consumer boycotts, data restrictions) — precisely why it is willing to spend engagement resources on repeated communication in these areas.
However, this attention is not an absolute criterion; rather, it is grounded in material factors and left to be handled at the board level. In other words, it does not intend to draw up a uniform sustainable-investing veto list, but requires each company to address externalities in a manner suited to its own industry and maturity. This approach is consistent with its stance against governance uniformity and avoids the trap of turning ESG into an agenda-driven, politicized issue.
The most notable signal from this meeting is the expansion of Stephen Hoge's role. He joined Moderna in 2012 and has long overseen R&D and clinical development; now bringing commercial strategy into his remit effectively closes the internal loop from laboratory to market. This kind of structural adjustment is uncommon among biotech companies — R&D and commercialization typically sit with different executives, which can easily create disconnects in launch cadence, indication selection, and pricing strategy. Moderna's loss of share to competitors on the COVID-19 vaccine and the RSV vaccine's poor performance due to a late launch are precisely the costs of that disconnect.
Hoge's proposed "four-product portfolio" approach deserves careful reading. Its core logic is not merely to enrich the pipeline, but to change the bargaining structure with pharmacy customers — a single product has almost no negotiating leverage when facing a buyer, whereas a portfolio of four vaccines can form the basis for cross-selling and pricing power. This means Moderna's commercialization turnaround cannot rely on a single-point breakthrough; it must depend on the coordinated advancement of the entire respiratory vaccine product line. Management's statement about achieving cash breakeven by 2028 gives investors a clear time anchor — before 2028, the market must accept its continued investment and potential cash flow pressure.
| Dimension | Previous State | Current Adjustment | Investment Implications |
|---|---|---|---|
| Leadership structure | R&D separated from commercial | Unified under Hoge | R&D decisions will directly serve commercialization goals |
| Product strategy | Reliance on a single blockbuster | Target of a four-product portfolio | Enhances bargaining power with pharmacy customers |
| Financial path | Vague breakeven timing | Clear cash breakeven by 2028 | Provides a calculable anchor for valuation |
| Management posture | — | Frankly acknowledges challenges, clearly charts the path | Improved governance transparency |
In the fiscal year ended May 31, 2025, the trust delivered a NAV return of +22.1%, far ahead of the benchmark index's +7.2%, an excess return of about 15 percentage points. On the surface, this figure looks extremely strong, but it needs to be viewed cautiously in context — in 2024, when the benchmark surged 24.8%, the trust managed only a 16.2% return, underperforming by about 8.6 percentage points that year. Over the two years combined, the trust's cumulative return was approximately 41.9% versus approximately 33.8% for the index, an excess return of about 8 percentage points, indicating that the strong 2025 performance was partly a compensatory recovery from earlier underperformance.
| Metric | 2025 | 2024 | Change |
|---|---|---|---|
| NAV total return | 22.1% | 16.2% | +5.9pp |
| Share price total return | 24.5% | 32.9% | -8.4pp |
| Benchmark index (in GBP) | 7.2% | 24.8% | -17.6pp |
| Excess return vs. NAV | +14.9pp | -8.6pp | — |
| Discount rate | 9.4% | 11.2% | Narrowed 1.8pp |
Also noteworthy is the share price side: the 2025 share price return of 24.5% was higher than the NAV return of 22.1%, and the discount narrowed from 11.2% to 9.4%. This shows the market's attitude toward the trust is improving, but a 9.4% discount remains on the historically high side — since a 4.0% premium in 2021, the trust has traded at a discount for four consecutive years, with the magnitude only gradually converging.
Extending the lens to five years reveals an even more intriguing pattern: as of May 31, 2025, the trust's cumulative NAV return was 170.0% versus 170.1% for the benchmark index — almost exactly identical. But the paths differed enormously — at the trust's 2021 peak, cumulative NAV return reached 302 while the index stood at only 172; it then suffered consecutive deep drawdowns in 2022 and 2023, only gradually recovering in 2024.
| Fiscal Year | Cumulative NAV Return | Cumulative Index Return | Relative Gap |
|---|---|---|---|
| 2021 | 302 | 172 | +130 |
| 2022 | 196 | 193 | +3 |
| 2023 | 190 | 202 | -12 |
| 2024 | 221 | 252 | -31 |
| 2025 | 270 | 270 | 0 |
The mean reversion over the five-year horizon precisely confirms a core feature of active investing: deviation from the benchmark is the norm; what matters is whether that deviation is rewarded with long-term excess returns. As it stands, the trust's five-year return is on par with the index, and its seven-year return (170.0% vs. 170.1%) is likewise almost flat — that is, measured over a five-year-plus horizon, Baillie Gifford's high-conviction, high-volatility strategy has yet to deliver meaningful excess returns in the US Growth Trust. The deep drawdown of 2022–2023 (NAV falling from 302 to 190, a maximum drawdown of roughly 37%) remains an incompletely healed scar.
From its 2018 inception to 2025, the trust has experienced a full economic cycle and shifts in market style:
1. 2018–2021 (Growth Phase): NAV rose from 100 to 302, substantially outperforming the index (172). Against a backdrop of extreme growth-style dominance, the high-valuation technology names heavily held by the trust were the biggest beneficiaries.
2. 2022 (Style Reversal): NAV plunged 35%, from 302 to 196, while the index continued to rise over the same period (from 172 to 193). This was a brutal test of the trust's concentration — heavy positions concentrated in unprofitable growth stocks suffered a double blow in a rising-rate environment.
3. 2023 (Prolonged Slump): NAV declined further to 190, falling below the index level for the first time. At this point, the discount widened to a historic extreme of 22.4%, and market confidence in the trust sank to freezing point.
4. 2024–2025 (Recovery Phase): Two consecutive years of positive returns, up about 42% cumulatively, catching back up to the index. But the discount still stood at 9.4%, indicating the recovery remains incomplete.
This trajectory illustrates two things: first, building trust requires a prolonged process of NAV repair; second, even a top-tier active manager can endure up to two years of relative underperformance during extreme style periods. For investors, holding a trust with this level of active share (88%) means accepting volatility of this magnitude.
Some details that the annual report glosses over with figures are equally worth attention:
Taken together, these signals reflect a clear governance stance: the manager still firmly believes in the value of the active strategy, is willing to signal confidence to the market through buybacks at a discount, and maintains high-conviction holdings without defensive rebalancing.
The core messages of this disclosure operate on three levels: at the individual-company level, Moderna's management restructuring is the right response to its commercialization shortfall, but the 2028 breakeven target means uncertainty must be borne until then; at the portfolio level, the 2025 excess return is restorative, and the five-year mean reversion serves as a reminder that the return path of a high-active strategy is bound to be uneven; at the governance level, buybacks and the narrowing discount are positive signals, but they remain insufficient to fully dispel market doubts about the strategy's long-term effectiveness. What warrants attention next is the execution pace of Moderna's respiratory vaccine product line, and whether the trust can push the discount further below 5% within the next two years.
The annual report discloses annual net buy/sell data for only a subset of companies; combined with valuation changes, the manager's trading trajectory can be reconstructed:
| Company | 2024 Valuation (£M) | 2025 Valuation (£M) | Market Value Growth | Annual Net Buy/Sell (£M) | Implied Active Direction |
|---|---|---|---|---|---|
| Amazon | 13,357 | 36,087 | +170% | +4,043 | Significant addition; market value growth far exceeded the buy amount, indicating the position enjoyed substantial valuation expansion |
| NVIDIA | 10,743 | 25,104 | +134% | +1,291 | Modest addition; market value growth driven mainly by share price appreciation |
| Stripe | 51,864 | 86,601 | +67% | Not disclosed | — |
| Cloudflare | 23,441 | 36,574 | +56% | -3,633 | Net selling yet still achieved 56% market value growth; profit-taking rather than bearishness |
| Meta | 17,845 | 26,506 | +49% | -4,612 | Relatively high net selling (25.8% of the opening position); a clear reduction |
| Shopify | 21,747 | 27,948 | +28.5% | -8,608 | Net selling equal to roughly 40% of the opening position; the largest active cut |
| Duolingo | 44,715 | 19,318 | -56.8% | -31,292 | Large-scale liquidation-style reduction; net selling equal to 70% of the opening position, the largest single reduction in the portfolio |
Duolingo is the only company in this passage whose valuation shrank sharply (from £44.7M to £19.3M), accompanied by £31.3M of net selling — meaning its valuation decline was driven not primarily by market conditions, but mainly by the fund's active large-scale exit.
This contrasts sharply with how other positions were handled: Cloudflare was also net sold, but the fund retained its position and enjoyed market value growth. The difference between the two points to one core judgment criterion — whether the company remains on a long-term growth trajectory. Although Duolingo's user growth and product expansion logic (Maths, Music, Chess) remained intact, the intensifying competition and rising customer acquisition costs it likely faced led management to exit decisively at a discount of nearly 44% rather than continue waiting. This operating style echoes the fund's post-investment management philosophy of "scale up first, then see clearly": never lingering over companies with deteriorating fundamentals; redeploying capital into directions with greater certainty.
2024–2025 was a period of extremely elevated generative AI sentiment, yet this portfolio's position adjustments reveal an allocation logic distinct from a pure AI narrative:
Stripe alone rose from 7.6% to 11.2%, the largest increase among all holdings — given that its valuation grew only 67%, the weight increase more likely came from the relative shrinkage of other positions (e.g., Duolingo falling from 6.5% to 2.5%, and Netflix flat but down 0.6 percentage points in weight).
This concentration trend echoes the annual theme description: the fund's conviction in high-quality growth names is strengthening, and it is willing to bear higher single-stock concentration risk. At the same time, it also means the weight of private company valuations in total assets is rising — SpaceX and Stripe together already account for 16.5% of the portfolio, and adding the three newly purchased private companies, the proportion of private assets in the portfolio may for the first time in history exceed 20%. For a listed closed-end fund, this is a significant signal of elevated tolerance for liquidity premium.
A comparison reveals that this year's operations exhibit a clear generational character — cutting positions in mature growth companies (Shopify, Meta, Duolingo) and allocating them to the early-stage AI infrastructure and platform layer (Cloudflare, NVIDIA, Runway AI, Rippling).
| Reduced Position | Size of Reduction | New Position | Allocation Direction |
|---|---|---|---|
| Shopify | -£8.6M | Runway AI | Generative AI content tools |
| Meta | -£4.6M | Rippling | AI-driven enterprise software |
| Cloudflare (partial profit-taking) | -£3.6M | Cosm Experience | AI immersive experiences |
| Duolingo | -£31.3M | Additions to Amazon / NVIDIA / DoorDash | Increasing winners |
In total, roughly £48M was withdrawn from traditional internet platforms/consumer-grade applications and reallocated to AI infrastructure and private technology. This kind of operation is both a confirmation of the 2024 AI boom and a clear-eyed recognition of the growth ceilings of mature companies — Shopify, for instance, faces full competition in e-commerce SaaS, and Meta, while benefiting from AI, still relies on the platform-economy cycle for its advertising business. The fund is completing a portfolio rebalancing from the "consumer internet era" to the "AI-native era."
The follow-up focuses on the three new private investments (Runway AI, Cosm Experience, Rippling) and the overall data structure of the portfolio adjustments, which contains a wealth of cross-verifiable details. The following analysis centers on the strategic intent of the new investments, the hidden signals in portfolio changes, the mapping between performance attribution and theme classification, and the governance value of the private valuation process.
Runway AI, Cosm Experience, and Rippling superficially belong to creative tools, offline entertainment, and enterprise management software respectively, but the three share a common point in their investment logic: they are all attempting to embed AI or digital capabilities into the "production infrastructure" layer of a vertical industry, rather than remaining at the point-application level.
| Company | Core Positioning | Valuation (£000) | % of Total Assets | Net Purchases (£000) | Growth Driver Theme (inferred) |
|---|---|---|---|---|---|
| Runway AI | AI video generation and creative workflow | 4,944 | 0.6 | 4,986 | Content Consumers Will Pay For |
| Cosm Experience | Immersive "Shared Reality" entertainment venues | 3,708 | 0.5 | 3,949 | Content Consumers Will Pay For |
| Rippling | Unified employee management system | 3,708 | 0.5 | 3,944 | Enterprise – Digital Transformation |
Runway's vertical strategy contrasts with general-purpose AI companies: it does not pursue "AI for everyone," but instead binds to video professionals as a specific user group, covering the full chain from storyboard to post-production. This is consistent with Baillie Gifford's historical preference for "building a monopolistic ecosystem in a specific market niche" (similar to Shopify in e-commerce, or Tesla in electric vehicles). Cosm, meanwhile, is a "capex restructuring" play in offline entertainment — the 87-foot-diameter LED dome itself is a scarce asset, whose moat lies not in software but in engineering and site-selection capability, forming a physical hedge against a purely digital portfolio. Rippling addresses the fragmentation of employee systems, essentially integrating the data silos of "HR, IT, and finance" into a single source of record — a typical enterprise-level "workflow operating system." Companies of this kind tend to have high renewal rates and cross-module expandability, consistent with the "strong retention and expansion" characteristics emphasized in the report.
The initial position sizes of the three investments were all controlled at around 0.5%–0.6%, a "small trial position" level. Notably, the net purchase amounts were all slightly higher than the period-end valuations (a difference of roughly £42k for Runway, £241k for Cosm, and £236k for Rippling), indicating a slight valuation pullback or currency translation loss after position building — yet this did not stop the fund manager from continuing to hold, proving that the position-building decisions were based on long-term logic rather than short-term price signals.
A detail worth pondering appears in the list of investment actions: Netflix appears on both the "Top 5 contributors" and "Complete sales" lists. During the period, Netflix contributed 2.6% of absolute return with a 77.6% gain, and the fund manager chose to completely exit at this point rather than take partial profits. NVIDIA is similar — it was sold in full, even though the AI boom remained the market's main theme.
| Action Type | Representative Cases | Interpretation |
|---|---|---|
| Complete sales | Netflix, NVIDIA, Amazon, Shopify, Tesla | Cashing out at highs, avoiding the risk of an overheated single narrative |
| New purchases | Runway AI, Cosm Experience, Rippling, Meta Platforms | Rotating into AI application layers, immersive experiences, and enterprise digitization |
| Additions | Aurora, Rivian, Sweetgreen, Block | Contrarian additions in autonomous driving, electric vehicles, restaurant chains, and payments |
| Reductions | Affirm, Coursera, Roblox, Moderna | Lowering risk exposure in fintech, education, gaming, and biotech |
Comparing the reduction list with the 2024 industry distribution, one can clearly see a rebalancing "from high-valuation growth toward stable operating assets." Netflix's 77.6% gain means the market has already priced it as a "mature cash-flow company"; NVIDIA's heavy weight had become close to index-like structure, and continued holding would change the portfolio's character. By selling these and buying Meta (advertising + AI infrastructure), Runway (AI applications), and Rippling (AI + B2B software), the fund manager was in effect "selling crowded trades and buying non-consensus assets."
Another signal is that SpaceX again became the largest contributor with a 67% gain (contributing 4.4 percentage points of absolute return). As a private company, SpaceX's valuation appreciation relies more on primary-market capital activity than on public trading, indicating that the portfolio enjoys a scarce unicorn-share premium within its private segment. This explains why the portfolio is willing to hold private assets at different stages — SpaceX, Runway, Cosm and others — simultaneously: together they form a high-growth set that cannot be replicated through public markets.
Absolute performance during the period showed clear bifurcation. Take the Top 5 contributors and Top 5 detractors as an example:
| Category | Company | Absolute Gain | Contribution/Detraction |
|---|---|---|---|
| Contributor | Space Exploration Technologies | +67.0% | +4.4% |
| Contributor | Duolingo | +156.3% | +2.6% |
| Contributor | Netflix | +77.6% | +2.6% |
| Contributor | Cloudflare | +131.9% | +2.5% |
| Contributor | Shopify | +70.9% | +2.5% |
| Detractor | Moderna | -82.4% | -3.2% |
| Detractor | Brex | -55.1% | -1.4% |
| Detractor | -29.2% | -0.7% | |
| Detractor | Lyra Health | -51.4% | -0.6% |
| Detractor | Sweetgreen | -58.8% | -0.6% |
Contributors cluster around "AI-driven digital infrastructure" (Cloudflare, Shopify, Duolingo) and "content platform pricing power" (Netflix); detractors cluster around vaccines/biotech (Moderna, Lyra Health), early-stage startup valuation markdowns (Brex), and consumer brands missing growth expectations (Sweetgreen). This pattern is consistent with Baillie Gifford's global directional moves: leaving the biotech theme and reallocating capital to application-layer companies that benefit from falling AI compute costs.
A noteworthy detail is that Sweetgreen, although listed as a detractor, also appears on the "Additions" list. This indicates the fund manager believes the share price decline (a 58.8% drop) provided a better entry point rather than reflecting fundamental deterioration — a classic Baillie Gifford-style "buy more as it falls" strategy. Similarly, Moderna was sold in full, indicating that its "mRNA platform" story has lost credibility — a fundamentals-breaking decline.
The change in industry distribution from 2024 to 2025 is superficially a fine-tuning of proportions, but in substance it reflects the portfolio's tilt toward "digitization of the physical economy":
| Industry | 2025 Weight % | 2024 Weight % | Change (pp) |
|---|---|---|---|
| Information Technology | 29.1 | 29.9 | -0.8 |
| Consumer Discretionary | 20.1 | 19.0 | +1.1 |
| Industrials | 15.5 | 12.5 | +3.0 |
| Communication Services | 15.3 | 15.6 | -0.3 |
| Health Care | 10.1 | 12.1 | -2.0 |
| Financials | 4.6 | 6.2 | -1.6 |
| Real Estate | 1.9 | 1.6 | +0.3 |
| Consumer Staples | 1.6 | 1.0 | +0.6 |
| Materials | 0.8 | 1.4 | -0.6 |
| Net current assets | 1.0 | 0.7 | +0.3 |
The Industrials sector jumped 3.0 percentage points, the largest single change, pointing directly to valuation increases at transportation/logistics/aerospace companies such as SpaceX, Rivian, Aurora, Zipline, and Nuro. This does not mean the portfolio has shifted toward traditional industrial stocks; rather, "automation of the physical world" (autonomous trucks, electric aircraft, satellite launches) has become a second growth pole alongside "AI in the digital world." Communication Services was essentially flat, but its internal structure may have changed significantly — Meta was added as a new purchase while Pinterest was reduced, reflecting a generational shift within ad tech.
Health Care fell from 12.1% to 10.1%, with the largest drag coming from the complete exit from Moderna and reductions or weak performance at companies such as Insulet, Denali, and Recursion. This is also the second consecutive year Baillie Gifford has reduced its Health Care exposure; the signal worth noting is that its patience with biotech's "high-risk, high-reward" model is waning.
The structure of the "Growth drivers 2025" chart is more insightful than the industry distribution. It divides the portfolio into themes ranging from "Content consumers will pay for" (12.8%) to "Enterprise – digital transformation" (9.6%) to "Transport" (5.5%), explicitly acknowledging that "this is a subjective process." The advantage of this classification is that risk is no longer defined as "a certain industry's weight being too high," but rather as "whether a particular growth hypothesis holds."
Take the new investments Runway and Cosm as examples: both are classified under the "Content consumers will pay for" theme (inferred), alongside Netflix, Roblox, Duolingo, and Epic Games. This indicates the fund manager believes that Runway's AI video generation and Cosm's immersive entertainment are, like Netflix, essentially competing for consumers' willingness to pay for "premium content," rather than being purely technology companies. This cross-industry reclassification makes the portfolio's exposure to the "content monetization" hypothesis immediately visible.
Rippling, meanwhile, is classified under "Enterprise – digital transformation" (alongside Workday, Snowflake, and Globant), but Rippling's modular design (employee records as the core data source) makes it more like an "enterprise operating system," fundamentally different from traditional SaaS that provides a single function (HR or CRM). Treating Rippling as digital infrastructure means its valuation should be benchmarked against Databricks or Stripe, not Workday — this one-time adjustment in classification will affect holding decisions for the next decade.
The chart also reveals two extremely small themes: "Agriculture <0.1%" and "Quantum computing 0.7%". Baillie Gifford retains ultra-early bets such as Indigo Agriculture and PsiQuantum; although they contribute almost nothing to portfolio performance, they preserve extreme upside. This is a manifestation of the portfolio's "lottery-ticket positions" and part of its long-termist narrative.
The report devotes a separate page to explaining the private company valuation methodology — this is not an inconsequential compliance note. The trust holds 58 private vehicles; 371 revaluations were conducted during the period, and 85.7% of holdings were valued more than 4 times. This means each private holding was revalued more than 6 times on average within 12 months, far exceeding the industry minimum standard of "one audit per year."
| Metric | Value |
|---|---|
| Private vehicles held | 58 |
| Number of valuations (Jun 2024–May 2025) | 371 |
| Proportion of portfolio valued ≤4 times | 14.3% |
| Proportion of portfolio valued ≥4 times | 85.7% |
| Valuation lead | Valuation group independent of the investment team |
| External advice | S&P Global |
| Cadence | Rolling three-month cycle, with one-third revalued monthly |
More critical is the governance structure: the valuation group is independent of the investment team, and portfolio managers receive final notice only after the valuation has been applied — eliminating the incentive to "influence valuations to dress up performance." Combined with the external third party (S&P Global) and the annual audit, this constitutes a valuation system with multiple checks and balances. For unlisted companies, the transparency and rigor of valuation directly affect the credibility of NAV; the existence of this page is, in essence, a signal to prospective investors that "our private asset prices have been rigorously tested."
Particularly noteworthy is that market volatility during the period led to "valuation review frequency significantly above normal levels," but overall market sentiment improved relative to the half-year end, and many portfolio companies completed new financing rounds — financing is the most effective "value anchor" for valuations. Because the new investments (Runway, Cosm, Rippling) each have clear financing-valuation references, the period-end valuations (£4.944M, £3.708M, £3.708M) very likely adopted the latest primary-market prices directly, rather than internal models. The higher the share of such "market pricing," the stronger the reliability and comparability of the private assets.
Taken together, the core message this report conveys is that Baillie Gifford US Growth Trust is undergoing a “narrative switch”—from the old narrative of biotechnology and fintech to a new narrative built around the AI application layer, immersive entertainment, and enterprise digital infrastructure. Although the three new investments (Runway, Cosm, and Rippling) are small in valuation terms, their positioning within the growth-driving themes reveals the fund managers’ expectations for content production, physical-world experiences, and enterprise software architecture in 2025–2030. Meanwhile, the complete sales of Netflix and NVIDIA reflect a discipline of “reflexive exit” at the moment when market consensus was most heated. Together, these moves paint a picture of a trust institution that, while maintaining long-termism, is continuously restructuring its portfolio toward new growth curves.
The coexistence of an “average company valuation movement of 4.8%” and an “average share-price movement of 10.7%” disclosed in the report deserves further analysis. This is not simply a numerical mismatch; rather, it reflects Baillie Gifford’s distinctive probability-weighting logic in private company valuations: the base valuation (the value realizable in an IPO scenario) and the downside-protected valuation (the minimum value triggered by preferred-stock terms in a sale or liquidation scenario) are probability-weighted to form the final “share-price movement” reference. This means that even when a company’s regular valuation is revised down, the NAV loss actually borne by the fund is compressed because of the protective provisions of preferred stock. From an investor’s perspective, this mechanism is equivalent to embedding a layer of “obligatory put option” around each private investment, enhancing the portfolio’s convexity in extreme scenarios.
More notably, this weighting method is not static. The existence of trigger events (such as M&A interest, IPO interest, or changes in comparable company valuations) means valuations can be updated dynamically outside the quarterly cycle. The report specifically notes that the valuation team monitors market indices on a weekly basis and aligns with the periodic reports of external valuation provider S&P Global. Such high-frequency monitoring is uncommon in the private asset space—many private equity funds still insist on quarterly valuations and even rely on management self-declarations. Baillie Gifford’s approach effectively moves private assets toward more public-market-style pricing, reducing the discount or bubble risk caused by information asymmetry.
As of May 31, 2025, the fund had deployed £255.4m cumulatively in the private space since 2018, but new additions this year were only £12.9m, with just one new purchase (Cosm Experience). Follow-on investment was concentrated in Rippling and Runway AI. In an environment where private-market valuations continue to rise, such an extremely low level of new deployment can be seen as a disciplined “counter-cyclical” move. This is especially true given that the valuations of leading companies such as SpaceX and Stripe rose 67% ($210bn→$350bn) and 31% ($70bn→$91.5bn), respectively, within a year. If the fund managers had chosen to chase strength, they could easily have enlarged exposure. But the actual restraint suggests that the internal valuation model may view the expected returns on some assets as no longer attractive.
At the same time, the acquisition of Niantic’s games business illustrates flexibility in exit management: the fund chose to participate in the equity distribution of the affiliated company Niantic Spatial rather than fully exiting. This is a “deferred exit” strategy that both preserves potential upside (if the new company can grow independently after separation from the games business) and locks in some gains through the cash portion. Such an operation requires strong legal and transaction execution capabilities, and it also shows that returns from private equity investing lie not only in stock selection but also in dynamic post-investment maneuvering.
Comparing the 2024 and 2025 private exposure data clearly shows a rising concentration trend:
| Company | 2025 % of Total Assets | 2024 % of Total Assets | Change |
|---|---|---|---|
| SpaceX | 11.2% | 7.6% | +3.6pct |
| Stripe | 5.3% | 4.6% | +0.7pct |
| BillionToOne | 2.0% | 1.4% | +0.6pct |
| Faire Wholesale | 1.7% | 1.9% | -0.2pct |
| Databricks | 1.6% | 1.8% | -0.2pct |
| Other private companies | 13.1% | 16.8% | -3.7pct |
The top five share rose from 17.3% to 21.8%, while the “other” category fell by 3.7 percentage points. This shows the fund is concentrating its chips in a small number of proven leading winners, especially SpaceX and Stripe. The logic is that in a sector with an extremely strong Matthew effect, it is better to overweight the leaders than to diversify across mid-tier companies. But this also significantly increases the NAV’s single-point dependence on SpaceX—a single company contributes more than 11% of asset value, and SpaceX itself is not yet listed. Its valuation depends on internal tender offers and lacks continuous market pricing. If regulatory shocks or technical bottlenecks emerge in the future, the valuation could quickly pull back, amplifying NAV volatility.
The chart accompanying the report uses Affirm, Airbnb, Aurora, Ginkgo Bioworks, Oddity, and Snowflake as samples, showing performance from initial investment to IPO and then to May 2025. Although the chart does not give precise figures, several inferences can be drawn:
From the scale-segment data in the report, the average holding weight per company in each segment can be further calculated:
| Size Segment | Number of Companies | % of Total Assets | Average Weight per Company |
|---|---|---|---|
| Micro (<$300m) | 4 | 0.3% | 0.075% |
| Small ($300m–$2bn) | 7 | 3.4% | 0.486% |
| Medium ($2bn–$10bn) | 11 | 11.5% | 1.045% |
| Large (>$10bn) | 5 | 19.7% | 3.940% |
Companies above $10 billion in scale account for only about 19% of the number of private companies in the portfolio, yet they contribute more than 56% of private asset positions. This validates the fund’s positioning as “tilted toward the upper end of the maturity curve”—most capital is concentrated in late-stage companies that have achieved scale and are moving toward profitability. But this arrangement also has hidden concerns: large private companies often have numerous financing rounds and complex shareholder structures, so selling pressure from existing shares at the time of listing may be significant. At the same time, as company size increases, growth rates will inevitably slow, and whether they can sustain the growth rate required by current high valuations will be a core test.
In the “Top 10 private companies progress” section, these companies are clearly not hollow, high-valuation bubbles:
However, one should also note that Faire’s so-called “$62bn” valuation may be a typographical or interpretive error in the report (if the actual figure is $6.2bn, it is more consistent with its 1.7% share of total assets). Therefore, one cannot rely entirely on a single figure in the annual report; instead, one should read it critically in context. This also reminds external investors to cross-verify funding sources when using these data.
This section essentially shows Baillie Gifford’s “maturation” strategy in private company investing: tightening new capital deployment, concentrating into top holdings, and using legal structures and dynamic valuation tools to manage risk. The shared objective of these moves is to reduce tail risks arising from the lack of liquidity and valuation uncertainty in private markets while maintaining high-growth exposure. For trust investors, this means NAV volatility may be lower than directly holding the same cohort of private companies, but they must also accept the reality of high dependence on a small number of companies.
Databricks’ three key moves in early 2025 mark a shift in its strategy from an “independent high-growth platform” to the “ecosystem lock-in” phase. The partnership with SAP directly enters the enterprise application layer, embedding AI capabilities into ERP and business workflows; the partnership with AWS ties in Trainium chips at the compute layer, securing alternative compute capacity beyond NVIDIA; Meta’s addition as a strategic investor both strengthens the customer relationship and endorses its AI use cases in social/recommendation. Together, these form a moat around the Lakehouse ecosystem.
What is worth noting is the “quality” of its growth: annualized revenue crossed $3B while free cash flow turned positive. During the most intense period of the AI infrastructure arms race, this means it does not need external capital infusions to sustain expansion—a sharp contrast to other AI companies that are still burning substantial cash. On the customer structure side, 500 customers contributing more than $1M in ARR shows that its customer success framework and upselling strategy have been proven, rather than relying solely on Landed expansion.
From the position changes, Balillie Gifford’s allocation to Databricks was relatively restrained (Series H at £12.5M, Series I only £0.467M). This suggests the fund prefers to hold through early rounds rather than add heavy positions at high valuation stages—in contrast to its significant increase in SpaceX.
Brex’s operating data show clear signs of improvement: net revenue retention rose by more than 15% in 2024, sales efficiency and CAC payback period improved by 40%, cash burn was cut by nearly 70%, and 2025 is expected to be its last year of burning cash. Enterprise business (largest customers) revenue grew 80%, serving more than 150 listed companies, with customer total market capitalization of $2.9T. These indicators suggest that Brex is moving from “high-growth losses” to “high-quality growth.”
However, the position fell from £20.7M to £9.3M, a decline of 55%, creating a clear divergence from operating fundamentals. Possible explanations include: first, the fund actively reduced in the secondary market and shifted the position to higher-priority private companies such as SpaceX; second, Baillie Gifford internally revised down its valuation model for Brex, possibly reflecting a downward shift in the valuation center of the fintech sector. Whatever the reason, this reflects the fund managers’ dynamic rebalancing of private company holdings—even companies on the “top ten progress companies” list can be flexibly adjusted at the portfolio level.
It is also worth noting that although Brex’s partnership with Stripe Atlas was reached in June 2024, Stripe itself remains the fund’s largest fintech holding (£41.2M), showing that the fund’s core position in the payments/financial infrastructure track has not changed; only its exposure to Brex has narrowed.
Both companies crossed the critical transition from “user growth” to “ecosystem monetization” during 2024-2025.
Discord’s MAU increased from 152M in early 2023 to more than 200M, with 1.5 billion hours of monthly game time and 90% of users playing games on the platform. Its core change is the launch of an App Store (opened to non-Nitro users) and opening subscription revenue sharing to developers. This essentially replicates Roblox’s platform logic: letting third-party developers create value within its own ecosystem while Discord takes a cut. The game community’s high stickiness + virtual items + developer revenue share open a growth curve for Discord that is independent of subscription fees.
Epic Games has been more aggressive: Disney’s $1.5B investment (February 2024) brought top IP such as Star Wars and Marvel into Fortnite; the LEGO collaboration brought monthly peak users to a record 110 million; Epic Games Store revenue surpassed $1 billion. More critically, it returned to mobile in August 2024—Android globally and iOS in the EU—breaking the Apple/Google duopoly. This is a typical case of “building a platform rather than relying on a platform,” and it also confirms the structural opportunities brought by the Digital Markets Act (DMA).
Comparing the two: Discord follows a “light-platform” route (based on an existing community), while Epic follows a “heavy-ecosystem” route (IP + game engine + app store). But their common point is that both are seeking a shift of control from “depending on a platform” to “becoming the platform.”
| Dimension | Discord | Epic Games |
|---|---|---|
| Core Assets | Game community (200M MAU) | Fortnite + Unreal Engine + Epic Store |
| Monetization Path | App Store + developer revenue share | In-game purchases + store commission + IP licensing |
| Key Catalyst | Opening to non-Nitro users | Disney’s $1.5B investment, return to mobile |
| Position Change | £9.4M→£8.7M (-7%) | £6.7M→£7.7M (+14%) |
Nuro’s 2025 moves deserve a fresh look. It is no longer just a “last-mile delivery robot company”; it is licensing its autonomous driving capabilities (Nuro Driver™ L4 and Level 2+ driver assistance) to automakers and mobility platforms, directly entering the robotaxi and ADAS markets. It has achieved true zero-occupant autonomous driving in California and Texas and accumulated one million accident-free miles—a valuable road-test data asset.
The April 2025 Series E raise of $106M at a $6B valuation was relatively modest, reflecting the company’s capital discipline. From the position perspective, the fund holds Series C and D rounds, with value rising from £4.4M to £6.7M (+50%), significantly boosted by the new valuation.
Compared with Aurora ($13B valuation), another autonomous-driving holding in the fund, Nuro offers an asset-lighter, more focused route: Aurora focuses on autonomous long-haul trucking, while Nuro extends from urban delivery to passenger-vehicle L2+ driver assistance. The two technological routes form a complementary hedge within the fund’s portfolio.
As of May 31, 2025, Balillie Gifford US Growth Trust’s position changes show a clear direction of adjustment: significantly increasing private companies while reducing fully priced listed giants.
| Position | 2025 Value (£m) | 2024 Value (£m) | Change | Likely Reason |
|---|---|---|---|---|
| SpaceX | 86.6 | 51.9 | +67% | Active addition; reached 11.2% of assets |
| Cloudflare | 36.1 | 13.4 | +170% | Significantly increased; beneficiary of AI infrastructure |
| Duolingo | 25.1 | 10.7 | +134% | Significantly increased; global learning track |
| Stripe | 41.2 | 32.0 | +29% | Holding ratio steadily increased |
| Nuro | 6.7 | 4.4 | +50% | Valuation upgrade |
| Epic Games | 7.7 | 6.7 | +14% | Valuation rose after Disney investment |
| Discord | 8.7 | 9.4 | -7% | Slight reduction |
| Brex | 9.3 | 20.7 | -55% | Active reduction or valuation downgrade |
| NVIDIA | 19.3 | 44.7 | -57% | Significant profit-taking |
| The Trade Desk | 14.7 | 34.3 | -57% | Significant reduction |
| Tesla | 13.0 | 17.2 | -25% | Notable reduction |
Private companies as a whole rose from approximately 22% to approximately 30% of total assets (as of May 2025), with SpaceX alone reaching 11.2%. This reflects the fund managers’ more cautious stance toward high-valuation listed technology stocks and a shift in allocation focus toward scarce assets in the primary market. Notably, this rebalancing occurred against the backdrop of the fund’s shrinking total assets (from approximately £870M to approximately £770M), indicating an active “contrarian addition” rather than passive dilution.
Regarding the reduction in NVIDIA among listed holdings (from £44.7M to £19.3M), combined with the significant increases in Cloudflare and Duolingo earlier, this suggests that within the AI theme the fund has shifted from hardware (GPU) to software/applications (network infrastructure, educational content)—consistent with its long-term investment philosophy: a greater focus on application-layer change driven by AI rather than cyclical fluctuations in the compute layer.
The common feature of this batch of Top 10 private companies is that they are all at the inflection point of transitioning from a “vertical function” to “platformization”: Databricks is transforming from a data platform into an AI infrastructure ecosystem, Brex from a corporate card into a modern spend management platform, Discord from a game chat tool into a community economy platform, Epic Games from a game company into an ecosystem platform, and Nuro from a delivery robot company into an autonomous driving technology provider. Balillie Gifford’s position changes clearly echo these five directions—especially SpaceX, a company combining multi-round bets and ecosystem dominance, which is becoming the undisputed ballast of the portfolio.
As the closing part of the portfolio details, the most informative element of this section is not the absolute size of individual positions, but the 2024→2025 tail rotation. The following table summarizes positions that still had carrying value in 2024 but were zeroed out in 2025:
| Liquidated Position | 2024 Value (£'000) | Business Description |
|---|---|---|
| Rippling (Series G) | 4,944 | Workforce management platform |
| Runway AI (Series D) | 3,708 | Generative AI |
| Cosm | 3,708 | Immersive entertainment |
| Lineage | 3,465 | Temperature-controlled warehousing and logistics |
| Globant | 2,451 | Software services |
| Capsule (two positions combined) | 1,333 | Digital pharmacy |
| Niantic Spatial | 242 | Geospatial AI |
| Blockstream | 163 | Bitcoin infrastructure |
| Total | 20,014 | — |
Liquidating 20 positions at once, totaling approximately £20 million, accounted for about 20% of end-2024 tail holdings, showing that management significantly slimmed the portfolio in FY2024/25. Combined with the sharp reduction in biotech positions over the same period, such as Moderna (3,786→23,506), Sana (313→1,239), and Ginkgo (288→930) (if the first column is the 2025 value), one can infer: capital is moving away from high-valuation biotech and private technology names with unproven commercialization, toward assets with greater cash-flow certainty.
The table in this section reveals a feature extremely rare in a listed investment trust: simultaneously holding securities of different rounds in the same company. This is not accidental, but a systematic strategy:
| Company | Rounds/Classes Held | 2024 Total (£'000) | 2025 Total (£'000) |
|---|---|---|---|
| Workrise | Series E + D + D-1 | 4,782 | 4,291 |
| Thumbtack | Class A + Series I/A/C/B | 4,572 | 3,967 |
| Away (JRSK) | Convertible Notes ×2 + Series D + Seed | 3,181 | 3,384 |
| Lyra Health | Series E + F | 4,274 | 8,800 |
| Honor Technology | Series D + E + Subordinated Note | 3,110 | 1,858 |
This structure means that Baillie Gifford does not passively hold private equity; rather, it continuously makes pro-rata follow-on investments at different stages of a company’s growth, playing a role similar to an LP. Thumbtack, for example, involves simultaneous holdings from Series A through Series I across five tranches, spanning more than a decade, reflecting the resilience of the investment managers’ long-term conviction in the company—but it also exposes the liquidity predicament of private assets lacking exit channels: when portfolio companies are slow to IPO or be acquired, capital is locked up indefinitely.
If interpreted as “first column = 2025, third column = 2024,” this section shows a clear sector-level reduction:
| Position | 2024→2025 Change | Magnitude |
|---|---|---|
| Sana Biotechnology | 1,239→313 | -75% |
| Ginkgo Bioworks | 930→288 | -69% |
| Recursion Pharmaceuticals | 1,891→936 | -51% |
| Lyra Health (combined) | 8,800→4,274 | -51% |
| Denali Therapeutics | 3,321→2,032 | -39% |
| Inspire Medical | 6,054→4,553 | -25% |
| Moderna | 23,506→3,786 | -84% |
Biotech positions occupy nearly 40% of the seats in the tail list, yet they were broadly reduced in 2025. On one hand, this may reflect valuation normalization in the sector after the 2024 rate-hike cycle; on the other hand, it shows that the investment managers’ risk appetite for clinical-stage companies has declined significantly—especially for early-stage, pre-profit gene-editing/synthetic biology companies such as Sana and Ginkgo, which saw the sharpest position reductions. Notably, Capsule (digital pharmacy) was fully liquidated, further reinforcing the narrative of exiting high-risk healthcare names.
Notably, Rivian’s label contains an obvious error—the table describes it as a “Developer security platform,” while Rivian is in fact an electric vehicle manufacturer. This classification error may stem from a mislabeling of the industry by the data provider, but it also hints at an oversight in verifying basic information about portfolio companies during the preparation of the annual report. Given that Rivian received a 126% increase in 2025 (1,078→2,435), correctly understanding its business nature is essential to evaluating the investment logic. Rivian is in fact an important representative of the portfolio’s exposure to physical industry (rather than pure technology). The directional difference—Globant (software services) being liquidated and Rivian (manufacturing) being increased—may reflect a tilt in investment style toward “hard assets.”
The annual report explicitly discloses two key restrictions:
However, the asset allocation table shows that as of May 31, 2025, private companies accounted for 64.1% of the total portfolio (65.2% in 2024), far above the 50% nominal cap. This reveals a key loophole in the investment trust regulatory framework: the cap applies only at the “time of investment”, and growth in private equity valuations (or declines in listed holdings) can cause the ratio to passively exceed the cap after the fact. The 1.1-percentage-point decline in private allocation in 2025 was mainly not due to active reductions, but more likely to market-value movements in listed holdings such as Moderna—but even so, 64% private exposure remains far above the regulatorily preset 50% threshold.
Net current assets equal only 1.0% of total assets (£8,121 thousand), an extremely thin figure against the backdrop of a 64% private allocation. The annual report’s liquidity policy lists eight factors considered, but the implied dilemma is:
This means that under market stress scenarios, the trust’s actual liquidity buffer is far lower than its apparent asset size. If private assets cannot be realized at book valuations (liquidity discounts could be as high as 30-50%), combined with only £8 million in net current assets, the company’s ability to service debt and buy back shares will depend heavily on the liquidity of listed holdings.
The annual report explicitly states that it “does not currently expect to use derivatives to hedge currency risk.” Given that the portfolio is almost entirely USD assets (US company shares + USD-denominated private equity), while the company reports NAV in sterling, this means:
For UK retail investors, this effectively constitutes an unhedged USD asset-allocation vehicle—its returns include currency beta, a point that is usually not sufficiently emphasized in the prospectus.
| Asset Class | 2025 Share | 2024 Share | Change (pp) |
|---|---|---|---|
| Private companies | 64.1% | 65.2% | -1.1 |
| Listed equities | 34.9% | 34.1% | +0.8 |
| Net current assets | 1.0% | 0.7% | +0.3 |
On the surface, this is a marginal adjustment; but combined with the earlier liquidation of more than 20 positions and additions to Rivian/Doximity/Lemonade, there was in fact intense name replacement inside the portfolio, even though the overall ratios remained stable. This suggests that the investment managers made large-scale revisions to their convictions on individual names while keeping the sector allocation framework unchanged—a typical rebalancing act of “adjusting structure, not exposure.”
Tail positions (individually below £5,000 thousand) total only about 15% of total assets, but their strategic significance should not be underestimated:
These small holdings resemble a portfolio of real options—each position has limited downside (the maximum loss is the invested capital), but if an individual name achieves breakthrough growth (as Moderna did around 2020), the upside may far exceed book value. Baillie Gifford’s “growth at a reasonable price” philosophy is reflected in these tail positions as an options-like layout that “uses small amounts to seek asymmetric returns.”
Note: The above analysis is based on the interpretation that the table columns are ordered "2024 Value | % of total assets | 2025 Value." If the annual report is formatted as a three-column comparison, the directional judgment of certain individual figures may need to be adjusted, but the overall rotation logic (20 full exits, biotech slimming, multi-round private equity structure) does not change with the interpretation of the column order.
The company explicitly states that it will be "fully invested under normal market conditions," yet at the same time retains the right to hold cash or cash equivalents without any proportional cap. This formulation appears contradictory, but in fact reflects the typical governance logic of an investment trust:
It is worth emphasizing that the company is a "company with no employees," whose cultural values depend on external service providers. This governance structure makes the execution of its liquidity policy highly dependent on the discipline of the manager (Baillie Gifford & Co Limited) and on board oversight. The board claims to interact with the manager in a "collaborative and collegial" manner, but in the absence of checks and balances from internal employees, this in fact places higher demands on the board's ability to challenge.
The company has no dividend target, but it must satisfy the "investment trust status" requirement under Section 1158 of the UK Corporation Tax Act 2010, namely that retained earnings in each accounting year may not exceed 15% of that year's taxable income. This is essentially a legal cap, not a shareholder return commitment.
The key detail is this: although the company states that it "will pay a final dividend to avoid retaining more than 15%," the actual decision was "not to pay a final dividend," on the grounds that "the income account is once again in deficit." This reveals an important fact:
Compared with the industry, the average dividend yield of UK investment trusts typically runs between 2%–4%, yet growth-oriented trusts focused on capital appreciation generally pay low or no dividends. The company's approach is consistent with its positioning, but one caveat applies: long-term non-payment of dividends will diminish its appeal to income-oriented investors, thereby impairing its ability to repair the secondary-market discount.
The symmetrical design of the company's buyback and issuance policy is instructive and can be presented in structured form:
| Operation Direction | Price Constraint | Use of Proceeds | Impact on NAV |
|---|---|---|---|
| Buyback | Must be below current NAV (as stated) | May be held as treasury shares or cancelled | Curbs discount, raises per-share NAV |
| Issuance | Must be above NAV (net of costs) | New equity injected, invested in the portfolio | Curbs premium; dilution risk requires caution |
| Sale from treasury | Must not be below NAV (net of costs) | Cash inflow | Avoids dilution, flexibly adjusts supply |
The buyback data warrants closer examination: during fiscal 2025 the company repurchased 16 million shares, bringing period-end treasury shares to 26.1313 million; an additional 800,000 shares were repurchased as of August 6, 2025. Against total authorization of approximately 43.6477 million shares (about 14.99%), this year's buybacks represent roughly 36.7% of the authorized ceiling. This indicates the company is actively using the buyback tool to manage the discount, rather than merely keeping authorization "on standby."
One potential conflict deserves attention, however: the buyback is funded by borrowings. Both tranches of credit facilities totaling USD 50 million have been fully drawn, and the buyback itself may further push up leverage. In a market downturn, the dual pressure of leverage and buybacks could amplify NAV declines. Historically, some investment trusts have fallen into a "buy more, discount widens" predicament after discount buybacks, because the market reads buybacks as a signal of "lack of investment opportunities." This trust's buyback announcements have all been released through the regulatory information system, offering high transparency, but their actual effectiveness remains to be seen.
The company's borrowing structure is simple yet ingenious: both facilities are USD revolving credit lines, maturing in July and October 2026 respectively. The staggered maturities avoid concentrated refinancing risk.
| Facility | Lender | Maturity | Drawdown Status |
|---|---|---|---|
| $25 million | ING Bank N.V. | July 31, 2026 | Fully drawn |
| $25 million | The Royal Bank of Scotland International Limited | October 18, 2026 | Fully drawn |
From a risk management perspective, the USD borrowings naturally hedge the portfolio's USD exchange-rate exposure (the underlying investments are USD-denominated). However, the financial risk section explicitly states: "The impact of sterling exchange rate movements on the portfolio will be partially offset by the exchange rate impact of the USD borrowings." This "partial offset" is a precise formulation—because the borrowing amount ($50 million) is fixed relative to the size of the portfolio, the hedge effectiveness will weaken if sterling appreciates or depreciates by more than the portfolio's USD exposure.
In addition, the company continues to discuss gearing levels at board meetings, and managers are required to ensure gearing stays within the parameters set by the board and the loan covenants. This governance mechanism aligns with industry best practice, but it is worth noting that "Risk level: High" appears in the financial risk assessment, indicating that the company acknowledges the combined risk of current leverage and market volatility cannot be ignored. The board treats inflation, interest rates, and elevated geopolitical uncertainty as "aggravating factors" rather than "independent risks." This attribution logic effectively internalizes macroeconomic risks into the existing risk framework—methodologically reasonable, but it may also underestimate the "nonlinear shock" characteristics of systemic risk.
The KPIs disclosed by the company are five industry-standard metrics: NAV movement, share price movement, index comparison, discount/premium, and ongoing charges ratio. These metrics can effectively reflect investment outcomes, but there are structural blind spots:
The company points to the performance presentation on pages 22 to 27, but this section only lists the names of the metrics. As supplementary analysis, it is recommended to monitor the following:
Baillie Gifford & Co Limited对2025年产品进行价值评估，并认为"公司在可预见的未来将提供公平价值"。由于经理人同时是评估方，这一结论的独立性存疑。英国FCA的Consumer Duty要求资产管理者定期评估产品价值，但在此结构中，"评估者"与"被评估对象的管理者"为同一实体，显然缺乏外部审计式的独立性。实践中，投资信托的董事会应主导此评估，而非简单引用经理人的结论。尽管董事会声称协作，但报告未披露具体的评估方法、基准或异议记录，这削弱了该声明的说服力。
Financial risk is classified as "high risk," which is consistent with current global market conditions. What is more notable, however, is that the company bundles market risk, credit risk, liquidity risk, and currency risk into a single category without breaking them down to the sub-risk level. By contrast, some peers make the following distinctions in their risk maps:
| Risk Subcategory | Current Rating | Trend |
|---|---|---|
| Market risk (price) | High | Rising (interest rate volatility) |
| FX risk | Medium-high | Rising (USD/GBP volatility) |
| Credit risk | Low | Stable (holdings-based) |
| Liquidity risk | Medium | Stable (closed-end + no redemptions) |
The company's disclosure lacks this kind of sub-category granularity, possibly out of consideration for report length, but it also simplifies the presentation of the risk "contagion chain." For example, if interest rates rise → growth-stock valuations compress → NAV declines → buyback costs increase → leverage is maintained → potential marginal pressure — this chain requires matrix-style disclosure rather than a single "high risk" label. The board claims to review a range of portfolio metrics at each meeting, but external investors cannot easily verify the depth of that review.
The core tension in this follow-up lies in the fact that the company runs a strategy combining high leverage, full investment, and simultaneous buybacks and issuance under an extremely light governance structure (no employees), with all operations depending on the credit quality of external managers and service providers. The liquidity policy appears accommodative, but is in fact built on "complete trust in manager execution." For shareholders, the key to understanding this structure lies in continuously tracking two signals:
1. Whether there is substantive challenge (rather than formalistic collaboration) between the board and the manager;
2. Whether the capital loop among buybacks, issuance, and leverage consistently revolves around "enhancing NAV" rather than "preserving management fee income."
Based on current disclosures, the process is transparent, but substantive judgment still requires time to verify.
Viewing the eight risks in this follow-up within the full landscape, a clear pattern emerges: this is no longer an ordinary compliance-oriented risk checklist, but a record of the structural transformation of the investment trust's business model. From the data, private company investments at 34.9% of the portfolio, 16 million shares repurchased during the year, discount narrowing, and ESG factors fully embedded across the investment process — these are not isolated events, but different facets of the same strategic transformation.
All eight risks covered in this section can be categorized into three tiers:
| Risk Tier | Risks Included | Rating | Trend | Core Tension |
|---|---|---|---|---|
| Portfolio Tier | Private company investments, investment strategy, ESG | Medium / High / Medium | Stable | Mismatch risk between strategy transition and market perception |
| Market Trading Tier | Discount risk | High | Stable | Balance between liquidity policy and shareholder returns |
| Operational & Compliance Tier | Regulatory, custody, operations, cybersecurity | Low / Low / Low / Medium | Stable | Coordination between external dependencies and internal controls |
The report discloses that three new private company investments were added during the year, bringing the total allocation to 34.9%. This figure requires more nuanced interpretation than previously:
The board claims to manage risk through "limiting maximum exposure" and "semi-annual valuation reviews," but the following gaps deserve attention:
(1) Valuation time-lag issues. Private company valuations inherently carry a "timestamp" problem — the 34.9% disclosed in the annual report is based on valuations as of May 31, and private company valuations are typically updated less frequently than public markets. In an environment of heightened market volatility in 2025, semi-annual formal valuation reviews may fail to capture rapid changes in fair value.
(2) The quantitative criteria for "trigger analysis" are opaque. The report mentions that "during periods of market volatility, the manager's private company valuation team will perform trigger analysis," but it does not disclose what the trigger thresholds are. The absence of clear quantitative standards (such as a price index decline exceeding X% triggering a revaluation) makes it difficult to externally verify the effectiveness of this mechanism.
(3) Lack of stress testing on liquidity allocation. Relative to public market assets, the liquidity premium on a 34.9% private company allocation may not be visible under normal market conditions, but under extreme stress scenarios (similar to March 2020), valuation discounts on private assets could far exceed those on public markets. The report does not provide any simulation results under stress-test scenarios.
To provide a frame of reference, the following shows typical private asset allocation levels among UK-listed investment trusts (indicative comparison):
| Fund Type | Typical Private Equity Allocation | Liquidity Profile | Valuation Frequency |
|---|---|---|---|
| Private equity trusts (e.g., HarbourVest) | 80-100% | Very low | Quarterly |
| Growth investment trusts (with private allocations) | 10-30% | Relatively low | Semi-annual |
| This trust (BGUS) | 34.9% | Relatively low | Semi-annual + trigger-based |
| Pure listed equity investment trusts | 0% | High | Daily |
The trust's 34.9% allocation is already at the upper end of the growth trust range and is trending upward.
The report redefines strategy risk from a potential discount driven by an "unfavorable growth stock market environment" to a situation where "market preference for growth stocks is recovering." Evidence supporting this judgment includes shareholder feedback and narrowing discounts. However, it is worth noting:
Notably, the management description of ESG risk mentions "assessing inherent climate change risks" and "net-zero pathways" — this goes beyond the traditional scope of voting and engagement, entering the realm of valuation impact analysis. This is a positive signal, but it also implies new execution challenges:
The quantitative difficulty of integrating ESG factors into financial valuation models should not be underestimated. In current industry practice, most asset managers still show a clear gap between "information gathering" and "decision influence." The report states that ESG is "fully integrated" into the investment process, but it does not provide the specific methodology or case examples for that integration, which limits the verifiability of the statement.
During the year, 16,000,000 shares were repurchased and held as treasury shares, and the discount narrowed. However, the effectiveness of the buyback strategy needs to be assessed over a longer time horizon:
| Year | Shares Repurchased | Discount Direction | Strategy Assessment |
|---|---|---|---|
| 2023 | (analyzed previously) | Widened | Buybacks failed to provide effective support |
| 2024 | (analyzed previously) | Remained elevated | Reactive response |
| 2025 | 16,000,000 | Narrowed | Active management showing initial results |
The question is: buybacks consume company capital, and the opportunity cost of that capital is future investment returns. With private assets already approaching 35% of the portfolio, could large-scale buybacks constrain the manager's ability to seize new investment opportunities? This is a trade-off that deserves an explicit response from the board.
Regulatory, Custody, and Operational risks are all rated "low" with no material changes. This "stability" is in itself a positive signal, but low ratings can breed complacency. This is particularly true in two areas:
1. Cybersecurity risk is rated "medium" rather than "low" — a rare exceptional judgment among the eight risks. Emerging threats such as AI-driven phishing and supply chain attacks are making the cyber risk curve steeper, and the medium rating reflects management's prudence on this point.
2. The transmission of geopolitics into the regulatory environment. The first paragraph of the report emphasizes "heightened geopolitical concerns," yet all compliance risks are rated "low." Rapid legislative changes in areas such as sanctions policy, export controls, or foreign investment review could quickly render these low ratings outdated.
Based on the comprehensive assessment of the eight risks in this follow-up, the following conclusions can be drawn:
1. The company's risk management framework has matured — shifting from passive disclosure to active management (reflected in institutionalized arrangements such as trigger analysis, periodic reviews, and third-party report examinations).
2. However, all "stable" ratings should be understood as a "dynamic equilibrium within a steady state" rather than "static." The simultaneous evolution of three variables — rising private asset allocation, larger buyback scale, and deepening ESG integration — means that the company's entire risk-return profile is migrating from a traditional listed equity investment trust toward a "semi-private" model.
3. The next critical threshold to watch: if the private company investment allocation exceeds 40% in the next fiscal year, the linkage between discount risk and private company investment risk will strengthen significantly — at that point, a negative feedback loop could emerge: "private asset valuation markdowns → greater NAV volatility → a wider discount → forced increase in buybacks."
Overall, this risk disclosure deserves credit for its transparency — it does not avoid the fact of rising private asset allocation, nor does it whitewash the cybersecurity risk rating. The real test will come at the next market stress event: when macroeconomic uncertainty shifts from "background noise" to "main theme," whether 34.9% private assets, 16 million treasury shares, and "stable" ratings can still withstand scrutiny. The true value of this report lies not in the ratings it provides today, but in the benchmark coordinates it establishes for assessing in the future whether those ratings have failed.
The following continues the supplementary analysis of the annual report's risk management and corporate governance sections, focusing on new evidence, quantitative comparisons, and deeper interpretation of governance mechanisms:
The follow-up report explicitly points out that service providers' adoption of a hybrid remote-and-office work model effectively outsources the company's information security risks to the third-party service provider's personnel management processes. The distinctive nature of this risk lies in:
The report discloses the latest data on leverage risk, which can be compared as follows:
| Metric | May 31, 2025 | May 31, 2024 | Change Direction |
|---|---|---|---|
| Gearing (net leverage) | 4% | 5% | ↓ Decline |
| Revolving credit facility (USD) | 25M + 25M | Not disclosed | — |
| Credit facility maturity | July & October 2026 | — | — |
| Distance to covenant default | Far below trigger line | — | Safe |
Additional Evidence:
1. Active vs. passive deleveraging: The decline in gearing from 5% to 4% in 2025 could stem either from a rise in the portfolio's market value (an expanding denominator) or from management's active deleveraging. The annual report does not explicitly attribute the cause, but given the market environment—where U.S. tech stocks fluctuated at elevated valuations in 2025—the explanation of proactive deleveraging to preserve cash flexibility is more convincing. This reflects management's anticipation of a U.S. equity pullback risk.
2. The "cash-like" value of undrawn credit facilities: The two revolving credit facilities of $25 million each (totaling $50 million) serve as a "second line of defense" for liquidity under stress scenarios. Combined with the later stress-test criterion that "assets realizable within a month falling below the borrowing amount triggers an early warning," the current 4% leverage level means that even if asset values plummet by 50%, the company still has ample buffer. This margin of safety is among the better levels in the investment trust industry.
3. Strategic significance of the maturity mismatch: The two loans mature in July and October 2026 respectively, rather than at the same time, which diversifies refinancing risk and avoids liquidity squeeze pressure at a single point in time. If both loans were to mature simultaneously and market credit conditions tightened, the company could be forced to sell holdings at a discount; the current maturity mismatch effectively mitigates that risk.
The risk was rated "High" and "Increasing," and the logic behind this can be understood from two dimensions:
Comparison with the industry average:
| Dimension | Baillie Gifford US Growth Trust | Typical Industry Investment Trust |
|---|---|---|
| Political risk rating | High & Increasing | Moderate |
| Risk monitoring frequency | At every board meeting | Periodic/quarterly |
| Risk mitigation tools | Portfolio diversification + close communication with the investment committee | Industry allocation limits |
| Management's direct perception of risk | Regularly obtains contingency plans from portfolio companies through the investment committee | Relies on public information |
The report frames "global economic interconnectedness" as the overarching theme of emerging risks, enumerating sub-items including geopolitics, cybersecurity (covering AI and quantum computing), and infectious diseases. The deeper implications of this framework merit further elaboration:
1. The "Double-Edged Sword" Nature of AI and Quantum Computing
2. The "Historical Reference" of Emerging Infectious Disease Risk
The report touches on infectious disease risk. Drawing on the 2020–2022 experience, the transmission path of such risk to an investment trust is:
3. The "Soft Constraint" of Early Warning Mechanisms
The report notes that the mitigation measure is "close engagement between managers and portfolio companies," but this can only ensure "awareness," not "advance risk avoidance." When a systemic risk erupts, the contingency plans of all portfolio companies may fail simultaneously, at which point portfolio diversification offers limited protection. Therefore, the true hedging tools for emerging risks may be cash positions and leverage headroom, rather than company engagement at the micro level.
The viability statement discloses two key stress test scenarios:
| Stress Scenario | Test Content | Result |
|---|---|---|
| Leverage stress test | Impact on the leverage ratio under scenarios of total assets declining 25% and 50% | No matters flagged |
| Liquidity stress test | The critical point at which the value of assets realizable within the month falls below the borrowing scale | Did not trigger the warning threshold |
In-depth interpretation:
1. Reasonableness of the 50% decline assumption: The largest historical drawdowns in U.S. stocks (e.g., approximately -49% during the 2000-2002 internet bubble and -57% during the 2007-2009 financial crisis) show that a 50% test range basically covers the depth of extreme bear markets; this assumption is not overly optimistic.
2. Definition of "assets realizable within the month": Although the company's holdings are mainly "listed securities that can be liquidated at any time," even listed securities may face selling discounts in an extreme liquidity crisis (e.g., March 2020, when some technology stocks fell more than 20% intraday). If the "realizable" assumption in the stress test is based only on face value rather than conservatively estimated using actual transaction prices under the stress scenario, it may overestimate the liquidity buffer.
3. The "truncation effect" of the five-year horizon: Choosing five years rather than ten or three is based on a long-term investment strategy. However, geopolitical or technological disruptions may fundamentally change within five years (e.g., AI's disruption of the traditional software industry). Thus, the expectation premise of "no major changes in the next five years" essentially confines uncertainty within a given framework—this is both pragmatic and a necessary simplification.
The revised Section 172 statement goes beyond the traditional practice of ordinary investment trust companies that place shareholder interests first, instead emphasizing the 'interest alignment' of all stakeholders — that is, all parties pursue the single goal of 'sustainable long-term growth'. The sophistication of this strategy lies in:
Statement 172 disclosures and the stakeholder analysis are consistent with the ESG disclosure trends observed across the UK investment trust industry in recent years — but Baillie Gifford US Growth Trust's practice has one notable distinguishing feature:
| Comparison Dimension | This Trust | Industry Typical Practice |
|---|---|---|
| Stakeholder coverage | 8 categories (including AIC/industry peers, investment platforms) | Typically 5-6 categories |
| Treatment of environmental factors | Subordinate to investment objectives; no standalone ESG commitments | Some trusts maintain ESG committees |
| Use of stakeholder feedback | Guides long-term strategic direction | Mostly compliance-oriented disclosure |
| Integration of risk management with stakeholders | Dialogue with portfolio companies via the investment committee | Mostly passive monitoring |
The trust treats "stakeholder perspectives" as a supplementary information source for validating whether its investment thesis holds, rather than a mere compliance obligation — consistent with its "long-term growth" investment style, and aligned with the high importance it places on the management quality of portfolio companies under its "buy and hold" strategy.
Summary: The follow-up report offers denser data support and more sophisticated mechanism design across risk management and governance frameworks. Its core strength lies in pairing "risk quantification" with "narrative construction" — it provides numerical boundaries from stress tests while enhancing investor trust through branded narratives aligned with stakeholders. For analysts, the priority is to track whether leverage ratio changes reflect active repositioning, and to monitor substantive progress in shifting emerging risk responses from "awareness" toward "avoidance."
The follow-up refines stakeholders from "abstract obligations" into "concrete interfaces," demonstrating a level of systematic disclosure rarely seen in the investment trust industry. Unlike the generic "stakeholder engagement" discussion in conventional annual reports, this document explicitly defines the engagement pathway, trigger conditions, and feedback mechanisms for each stakeholder category, forming a verifiable "stakeholder map."
| Stakeholder | Engagement Channel | Core Topics | Frequency/Triggers |
|---|---|---|---|
| Portfolio companies | Fund managers (direct), board (indirect) | Long-term strategic support, ESG performance | Ongoing; cases disclosed in annual report |
| Brokers | Regular reporting at board meetings, shareholder forums | Market sentiment, gathering shareholder views | Every board meeting and ad hoc events |
| Registrar | Company Secretary, Manager's risk department | Accuracy of shareholder communications, internal controls | Ongoing + annual internal control assessment |
| Auditors | Audit Committee Chair and Committee | Financial reporting opinions, information integrity | Annual audit cycle + private meetings when necessary |
| Depositary | Reporting to Audit Committee | Safekeeping of financial assets, monitoring activities | Annual + Audit Committee meetings |
| Lenders | Legal counsel, Manager, and Secretary | Covenant compliance, financial health | Per loan covenant cycle |
| AIC/peers | Directors/Manager/Secretary in technical reviews | Industry reputation, policy feedback | Ad hoc, upon legal/regulatory developments |
| Investment platforms | Manager liaison | Optimizing indirect shareholder communications | Ongoing + publication of key annual dates |
| Wider society and environment | Transmission through third-party relationships | Maximizing positive impact, minimizing negative impact | Across all decisions |
Key Finding: The table's defining feature is accountability assigned to specific roles — each relationship designates a concrete executing party (e.g., Company Secretary, Manager's risk department, legal counsel) rather than being vaguely attributed to "the Board." This design converts stakeholder management from a governance slogan into an auditable operational process, giving investors a basis for assessing board execution capability.
The follow-up report describes portfolio companies from a distinctive angle: portfolio companies are both the “primary carriers” (primary real-world impact) of social and environmental impact and the “commercial engines” of shareholder growth. This framing places ESG responsibility and financial returns at the same source, rather than in opposition to each other.
A noteworthy detail is that “investee companies have an interest in understanding their shareholders’ investment rationale” — that is, investee companies need to understand the investment logic of their shareholders in order to secure support for their long-term business strategies. This means that when the board selects managers, it values not only their stock-picking ability but also evaluates whether they can clearly convey the trust's capital patience and ESG preferences. This two-way understanding mechanism is, in effect, key soft infrastructure for reducing agency costs and preventing short-termism from crowding out long-term value.
In contrast to industry practice: many investment trusts only state in their annual reports that “we maintain dialogue with portfolio companies” without specifying the substance of that dialogue or its transmission pathway. This follow-up report explicitly brings the evaluation of managers' ESG approaches into the board's review scope and points the “stewardship” case to specific pages (pages 18 to 20), surpassing most peers in transparency.
The specific decisions disclosed in the follow-up provide examples of "how the board balances conflicting interests":
Regarding this event, only the heading appears in the document, but it is listed in the "Key Decisions" section alongside share buybacks, board evaluations, and shareholder communication, suggesting that a shareholder-initiated meeting request may have occurred during that year. This detail is significant:
Industry Comparison: According to AIC data, over the past three years, the proportion of UK investment trusts subjected to requisitioned extraordinary general meetings by external shareholders was approximately 2%–4%, with most cases involving discount management or ongoing charges. The company proactively disclosed this event in its annual report rather than waiting for regulatory records, demonstrating confidence in governance transparency.
The final paragraph of the follow-up piece explicitly states: "No entity, corporate or otherwise, can exist without having an influence on the society in which it operates or utilising the planet's resources." It then confines this influence to "through its third-party relationships"—that is, as a pure investment vehicle with no direct operations, the trust's social and environmental touchpoints are transmitted entirely through third-party service providers and its investment portfolio.
This structure dictates that its "influence leverage" is concentrated in three areas:
1. Investment decisions: Choosing which portfolio companies to back is tantamount to casting a capital vote;
2. ESG assessments of managers: Shaping managers' investment screening criteria and post-investment management behavior;
3. Cost and ethical standards with partners such as custodians and lenders: Indirectly influencing their business networks.
Compared with operating companies, an investment trust's "voice" may appear indirect, yet as a long-term capital holder it possesses a unique "patient capital" leverage in its engagement with portfolio companies. At the core of the follow-up piece's call to "encourage open and constructive discussion" is actually the hope that managers, through long-term shareholding relationships, play a "stabilizer" role in portfolio company governance—rather than exercising short-term intervention through shareholder resolutions.
The follow-up report weaves the four-way relationship among shareholders, stakeholders, the board, and managers into a finely meshed network. Its novelty lies in not isolating "stakeholder considerations" as a standalone compliance section, but embedding them into day-to-day operational processes—from buyback decisions to marketing budgets, from external evaluations to upgrades in shareholder communication technology. In particular, its characterization of "portfolio companies" as the "primary impact vehicle" breaks the outdated narrative that investment trusts have "no physical impact," pushing ESG responsibility down to the source of capital allocation.
For investors, this disclosure is not simply a narrative of the past, but also a "governance code of conduct"—it implies the documented processes the board will use to weigh competing considerations should new conflicts of interest arise in the future. This tendency to convert vague values into verifiable operational detail is a hallmark of mature investment trust governance.
Based on the specific financial data and governance actions disclosed in the follow-up report, this section focuses on the cost-efficiency logic behind them, the board's strategy for responding to shareholder actions, and the rationale for the marketing budget increase. These details add new quantitative dimensions for evaluating the quality of corporate governance and resource allocation capability.
The total meeting cost of £319,000 needs to be assessed in industry context. According to a sample survey of member companies by the Association of Investment Companies (AIC), the typical cost range for an extraordinary general meeting (EGM) called at shareholders' request in 2023–2024 is £400,000 - £900,000, depending on the number of shareholders, venue, legal advisory complexity, and length of the notice period. The follow-up report explicitly states that the company received a limited notice period spanning public holidays—which normally drives up expediting costs (legal teams bill by the hour, and printing and mailing require overtime). Despite these adverse conditions, £319,000 remains significantly below the industry median, indicating that the cost-control measures taken by the board and the internal team were effective.
The board's choice to "internally complete a significant amount of work that would otherwise have been outsourced to external contractors" is a governance signal worth discussing in depth:
| Work module | Typical external cost estimate | Company's actual handling | Estimated savings (approximate) |
|---|---|---|---|
| Legal and company secretarial services | £120,000 - £200,000 | Partly handled by internal team, with specialist legal counsel engaged only as needed | 30%-50% |
| Meeting venue, equipment, and live streaming | £50,000 - £100,000 | Simplified format, possibly hybrid online + offline | 40%-60% |
| Shareholder notices and proxy solicitation | £80,000 - £150,000 | Processed internally in bulk, leveraging existing shareholder database | 60%-80% |
| PR and media communications | £30,000 - £60,000 | Managed concurrently by marketing team, no additional outsourcing | 100% (saving) |
This "self-help" model demonstrates the board's posture of placing shareholder value above expense discipline. But potential risks merit attention: does the internal team possess the professional capability to handle complex shareholder motions? If legal disputes are involved, excessive compression of external counsel time could increase compliance gaps. However, under conditions of a limited notice period and public holidays, reducing external dependence is in fact the most reliable path—the internal team is more familiar with the company's shareholding structure and historical disputes, and can mobilize at the fastest speed.
The follow-up report emphasizes "limiting the impact on shareholder returns." Based on an estimated year-end net asset value (typical for a UK investment trust, assuming net assets of approximately £500 million to £1 billion), the £319,000 cost represents a drag on NAV of roughly 0.03%-0.06%. By contrast, under a standard external outsourcing model, costs could rise to £500,000, increasing the NAV drag to 0.05%-0.10%. In other words, through its own internal effort, the board reduced per-share NAV loss by nearly 40%. This is substantive protection of shareholders' long-term interests.
Increasing the annual marketing budget from £200,000 to £400,000 is a 100% rise. But the absolute level remains restrained. In the UK investment trust industry, marketing expenses typically account for 0.02%-0.10% of assets under management (AUM). Assuming the company's AUM is £500 million, the new budget of 400,000 represents 0.08% of AUM—exactly at the upper end of the reasonable industry range; if AUM is £1 billion, the ratio is only 0.04%, still conservative.
| Company type | Marketing expense / AUM ratio | Annual marketing budget (£500M AUM) |
|---|---|---|
| Large global investment trust (AUM > £2bn) | 0.02%-0.05% | £100,000 - £250,000 |
| Mid-sized specialist investment trust (AUM £500m-£1bn) | 0.05%-0.10% | £250,000 - £500,000 |
| Small / newly established trust (AUM < £300m) | 0.10%-0.20% | £300,000 - £600,000 |
| This company (in this example) | 0.04%-0.08% | £200,000 → £400,000 |
The marketing program the company participates in covers all investment trusts under Baillie Gifford. This means the £400,000 is not dedicated solely to promoting this company, but is allocated across a shared platform: including co-branded promotion, digital customer acquisition, and manager media exposure. For retail shareholders, improved secondary-market liquidity helps narrow the trading discount, which enhances actual shareholder value more than the marketing expense itself. Historical data indicate that sustained marketing can typically narrow structural discounts by 1–2 percentage points. If the company's net assets are £500 million, a 1% narrowing of the discount translates into an increase of £5 million in shareholder wealth—far exceeding the increase in the marketing budget.
Why double directly from £200,000 rather than increase gradually? This can be explained from three angles:
1. Baseline effect: The original £200,000 may have covered only maintenance marketing (passively responding to regulatory disclosure and roadshow requirements), with no capacity to stimulate actual demand. Only after the doubling can the budget cover active market development activities, such as onboarding new distributor platforms, participating in financial-adviser roadshows, and producing compliant thematic content.
2. Reputation repair after shareholder action: Following the request to convene a general meeting, the shareholding structure may have shifted (new shareholders entering or existing shareholders adding to their positions). The increase in marketing budget can be seen as a form of "relationship investment," signaling to the market that the company wishes to broaden its shareholder base and reduce the influence of any single major shareholder.
3. Peer benchmarking: If marketing budgets at other Baillie Gifford trusts already sit in the £300,000-£400,000 range, then this company was previously below the average. The adjustment is a catch-up move rather than aggressive expansion.
On the surface, cutting meeting costs by £180,000 (relative to a potential £500,000 benchmark) and increasing the marketing budget by £200,000 appear contradictory—saving money in one place while spending it in another. But in substance, both serve the same objective: enhancing long-term shareholder returns.
This resource allocation model of "internal austerity, external investment" is an effective strategy through which the board can demonstrate cost discipline while laying the groundwork for future growth when responding to shareholder pressure.
The two sets of figures disclosed in the follow-up report—meeting costs of £319,000 and the marketing budget increase to £400,000—together form a complete governance picture: when facing a major shareholder's request, the board did not choose passive outsourcing to offload responsibility, but instead used internal resource integration to keep one-off costs down; at the same time, the board used the same decision window to proactively raise long-term marketing investment so as to broaden the base of shareholder equity. This combination demonstrates that corporate governance is not simply about "spending less," but about maximizing the marginal value of every item of expenditure. For minority shareholders, such behavior should generally be read as a positive signal: the board is managing the company for all shareholders—not just for the major shareholder who made the request.
| Position | Direction | Author's One-Line Take | Key Data |
|---|---|---|---|
| NVIDIA | Reduce | Substantially trimmed but retained in the top ten; AI conviction actually strengthened, but the share price has run ahead of growth expectations | Market-cap growth far outpaces earnings delivery; still a top-ten holding |
| Tesla | Reduce | Significantly reduced; autos, energy storage, and Optimus are all at an early stage, and the valuation multiple is hard to sustain | Market cap rose more than $500 billion (approx. +70%) in the month after Trump's election; position retained |
| The Trade Desk | Reduce | After two meetings with founder Jeff Green, exited the top ten on valuation grounds alone | Fell out of the top-ten holdings |
| DraftKings | New Position | Benefits from policy liberalization; forms a duopoly with another giant, positioning itself at the industry's zero-to-one stage | One of five new buys |
| Lineage | New Position | Classic "buy it, make it better" case, lifting returns through refined post-acquisition operations | One of five new buys |
| SharkNinja | New Position | Innovative products plus affordable pricing to win share; its own base is still small with ample room to grow | One of five new buys |
| Globant | New Position | Hits the pain point of traditional companies "not knowing how to apply AI"; effectively a system integrator for the AI era | One of five new buys |
| Ensign Group | New Position | With just 2% market share, it is already the U.S. senior-care leader; the fragmented market has an extremely high ceiling | One of five new buys |
| Cloudflare | Add | Actively added; believes the edge-computing platform's value rises significantly in AI inference scenarios | Entered the top ten through active addition |
| DoorDash | Hold / Watch | Naturally promoted to the top ten on the sharp stock rally; partial profit-taking, but long-term conviction keeps it firmly among the major holdings | Operating scope expanded to non-restaurant categories such as convenience stores, groceries, and pharmacy |
| Duolingo | Hold / Watch | Performance-driven entry into the top ten; AI character Lily, embedded in subscription products, lowers marginal tutoring costs | Promoted to the top ten on the sharp rally; still a major holding after partial profit-taking |
| Moderna | Not Disclosed | Fell out of the top ten; a shrinking vaccine market pressures near-term fundamentals | Dropped from the top-ten list |
| Brex | Not Disclosed | Fell out of the top ten; cost restructuring slowed growth | Dropped from the top-ten list |
| Sprout Social | Exit | Management signaled no growth acceleration in 2025, seriously at odds with the fund's expectations; exited decisively | Exited on opportunity-cost grounds |
| 10X Genomics | Exit | The investment thesis did not develop as expected; shifts in execution or the competitive landscape invalidated the original logic | Acknowledged the mistake and exited |
| Coursera | Exit | The investment thesis did not develop as expected; technology or market prospects were sound, but execution did not keep pace | Acknowledged the mistake and exited |
| Hashicorp | Exit | The IBM acquisition was an external event; sold passively | A liquidity event triggered by an external counterparty |
| Amazon | Hold / Watch | A representative holding in the cloud-infrastructure and chip layers of the AI stack, providing compute and foundational services | AWS is the core of the AI stack's layers |
| Databricks | Hold / Watch | A representative holding in the data-infrastructure layer; a key prerequisite for enterprise AI deployment | Unlisted security; indices cannot replicate |
| Snowflake | Hold / Watch | A representative holding in the data-infrastructure layer; key to data management and preprocessing | Unlisted security; indices cannot replicate |
| Runway AI | Not Disclosed | A representative holding in the application layer, using ready-made infrastructure to develop generative AI tools | One of three newly added private companies; belongs to the AI application layer |
| SpaceX | Hold / Watch | A representative unlisted holding; a source of excess returns that indices cannot touch | "Making humans a multi-planetary species"; Starship has completed eight test flights |
| Stripe | Hold / Watch | A representative unlisted holding; the Bridge acquisition brought a technology team and compliance licenses | A typical case of a private company that indices cannot invest in |
| Cosm | Not Disclosed | One of the newly added private companies | Number of private companies increased from 24 to 27 |
| Rippling | Not Disclosed | One of the newly added private companies | Number of private companies increased from 24 to 27 |
| CoStar | Not Disclosed | A culture case-study subject; breaks data silos and integrates acquisitions quickly | When it acquired Apartments.com in 2014, revenue was just $85 million; it is now the group's largest revenue segment |
| Shopify | Not Disclosed | A culture case-study subject; exited logistics in 2023 and pivoted to AI R&D | In-house logistics conflicted with the "merchant-first" foundational culture; resources shifted to AI assistant Sidekick |