Fundsmith Annual Letter to Shareholders 2024

Theme and Background

This chapter is the annual letter from Fundsmith Equity Fund to investors for January 2025, reviewing the fund's 2024 performance and elaborating on its investment strategy. The report notes that market returns in 2024 were highly concentrated in a few technology stocks, and the fund underperformed its benchmark due to insufficient holdings in these stocks. However, its long-term performance remains significantly superior to both the index and its peers.

Core Thesis

The author's core investment argument is: Short-term underperformance relative to the market is normal and should not alter the strategy of holding high-quality companies for the long term. Counter-intuitive judgments include:

• The fund returned +8.9% in 2024, underperforming the MSCI World Index's +20.8%, but its annualized return since inception is 2.7% higher than the index, with lower downside volatility (Sortino Ratio 0.87 vs 0.60).
• Market concentration is extremely high: Nvidia alone contributed over 20% of the S&P 500's return; 41% of the German DAX Index's return came from SAP (P/E ratio of 97x). The author believes this concentration is unsustainable and that the fund's decision not to chase weight-based allocations is reasonable.
• Among the five worst-performing stocks (L'Oréal, IDEXX, Nike, Brown-Forman, Novo Nordisk), most are considered by the author as "too good a company to sell," preferring to endure short-term pain.

Key Arguments and Data

• Fund Long-Term Performance: From its inception in November 2010 to the end of 2024, cumulative return was +607.3%, annualized +14.8%; ranked second among 162 funds in the Investment Association Global sector, outperforming the industry average (254%) by 353 percentage points.
• Market Concentration:

Market/Index	Concentration Data
S&P 500 (2024)	5 stocks (Nvidia, Apple, Meta, Microsoft, Amazon) contributed 45% of returns; Nvidia alone contributed over 20%
DAX Index (Germany, 2024)	SAP alone contributed 41% of returns, with a stock price increase of 69% and a P/E ratio of 97x

• Fund Performance Attribution:
• Top 5 Contributors: Meta Platforms (+4.1%), Microsoft (+1.6%), Philip Morris (+1.5%), Automatic Data Processing (+1.3%), Stryker (+1.3%).
• Top 5 Detractors: L'Oréal (-2.0%), IDEXX (-1.2%), Nike (-0.7%), Brown-Forman (-0.6%), Novo Nordisk (-0.6%).
• Stock Details:
• L'Oréal was dragged down by China's real estate and credit issues, but the author believes its business fundamentals are sound.
• IDEXX faced pressure from a post-pandemic slowdown in pet veterinary visits, but the author views it as an industry leader with clear long-term growth prospects.
• Nike suffered from increased competition due to management neglecting traditional retail channels, but a CEO change in 2024 led the author to adopt a wait-and-see approach.
• Brown-Forman was impacted by post-pandemic consumption declines and weight-loss drugs, but the author emphasizes its family control, high proportion of premium spirits, and its survival of Prohibition.
• Novo Nordisk's stock fell 10%, with a P/E ratio half that of competitor Eli Lilly, but revenue grew 20% annually. The author believes its market leadership and capacity advantages are solid.

Companies/Assets Involved

Fundsmith Equity Fund returned +8.9% in 2024, with a cumulative return of +607.3% and an annualized return of +14.8% since inception. The Sortino ratio of 0.87 significantly outperforms the MSCI World Index's 0.60

• Meta Platforms: Top 5 contributor (+4.1%), the author self-deprecatingly notes "the most criticized stock each year ends up performing the best."
• Microsoft: Top 5 contributor (+1.6%), purchased at approximately $25 in 2011, closing at $422 at the end of 2024, appearing in the top five for the ninth time.
• Philip Morris: Top 5 contributor (+1.5%), appearing for the fourth time, benefiting from reduced-risk products (e.g., heated tobacco) and the acquisition of Swedish Match.
• Automatic Data Processing (ADP): Top 5 contributor (+1.3%), appearing for the second time, with stable performance.
• Stryker: Top 5 contributor (+1.3%), appearing for the fifth time, benefiting from the recovery of elective surgeries delayed during the pandemic.
• L'Oréal: Detractor (-2.0%), the author is bullish, viewing China's issues as temporary.
• IDEXX: Detractor (-1.2%), the author is bullish, emphasizing its industry leadership.
• Nike: Detractor (-0.7%), the author is neutral to cautious, awaiting new management.
• Brown-Forman: Detractor (-0.6%), the author is bullish, emphasizing historical resilience.
• Novo Nordisk: Detractor (-0.6%), the author is bullish, believing the valuation is undervalued.
• Diageo: Mentioned as sold, but no specific data provided.

Investment Implications

• Do Not Chase Market Hype: The author explicitly refuses to allocate to tech stocks based on index weights, believing concentration risk is too high. Investors should be wary of the fragility caused by excessive concentration in a single stock or sector.
• Adhere to "Buy Good Companies, Don't Pay Too High a Price, and Do Nothing": The fund's strategy emphasizes long-term holding, not selling easily even during short-term underperformance. For "old friends" like L'Oréal, IDEXX, and Novo Nordisk, the author advises patience for a rebound.
• Monitor Management Change Signals: Nike's CEO change is seen as a positive signal. The author suggests that US companies hold executives accountable more quickly (contrasting with Unilever's 20% stock price increase), and investors can look for similar governance improvement opportunities.
• Beware of the Long-Term Impact of Weight-Loss Drugs on Consumer Stocks: Brown-Forman and Diageo are both affected, but the author believes premiumization ("drink less but better") may hedge the risk.

New Arguments and Data Analysis: Deep Insights into the 2024 Portfolio

1. Portfolio Quality: Sustained Advantage Over the Index

Although the portfolio's cash conversion rate fell from 91% in 2023 to 85% in 2024, below historical levels (~100%), other core metrics still significantly outperformed market benchmarks. Specific comparison data is as follows:

Metric	Fundsmith 2024	S&P 500 2024	FTSE 100 2024
ROCE	32%	16%	17%
Gross Margin	64%	45%	42%
Operating Margin	30%	16%	15%
Cash Conversion	85%	85%	90%
Interest Coverage	27x	9x	9x

Key Findings:

• ROCE and Margins: Fundsmith's ROCE (32%) is double that of the S&P 500 (16%), and its operating margin (30%) is nearly double the market, demonstrating sustained leadership in capital efficiency and pricing power.
• Attribution of Cash Conversion Decline: The 2024 decline was primarily driven by a surge in capital expenditures from four companies (Alphabet, Microsoft, Meta, Novo Nordisk). Novo Nordisk spent €10 billion to acquire three production sites to expand Wegovy capacity; the tech giants competed in GPU chips and data center construction. This reflects the pressure of the "AI arms race" on short-term cash flow, but Novo's spending is based on proven demand and competitive advantages, making the risk relatively controllable.

Among the five worst-performing detractor stocks in 2024, L'Oréal contributed -2.0%, IDEXX contributed -1.2%, and Nike, Brown-Forman, and Novo Nordisk each contributed approximately -0.6%

2. Valuation Analysis: Is the Premium Justified?

• FCF Yield Comparison: The weighted average FCF yield of the Fundsmith portfolio edged up from 3.0% at the start of 2024 to 3.1% at year-end, while the S&P 500 median FCF yield was 3.7%. Although the portfolio trades at a higher valuation (a premium of ~19%), given its significant advantages in ROCE (32% vs 16%) and gross margin (64% vs 45%), this premium is not unreasonable.
• 2025 Expectations: If the cash conversion rate reverts to historical levels (~100%), the portfolio's FCF yield could rise to around 3.6%, narrowing the gap with the S&P 500. Historical data shows that when cash conversion improves, the valuation premium often narrows (e.g., from 2017-2019, the portfolio's average FCF yield was 3.4%, with a smaller gap to the S&P 500 median of 3.8% than currently).

3. Transaction Costs and Quantitative Validation of the "Do Nothing" Strategy

• Extremely Low Transaction Costs: The portfolio turnover rate in 2024 was only 3.2%, with voluntary transaction costs accounting for just 0.002% (0.2 basis points) of the fund's average net asset value. In contrast, the average turnover rate for actively managed funds is typically 50%-100%, corresponding to transaction costs of 0.5%-1.0%.
• TCI vs OCF: Fundsmith's TCI (1.05%) is only 0.01% higher than its OCF (1.04%), far below the industry average. For example, the average TCI for UK equity funds is typically 0.3%-0.5% higher than the OCF, mainly due to commissions and spread costs from high turnover.
• Long-Term Holding Record: Since inception in 2010, 4 companies in the portfolio have never been sold, 9 have been held for over 10 years, and 15 for over 5 years. This "buy and hold" strategy not only reduces costs but also avoids the tax and opportunity costs associated with frequent trading.

4. Logic Behind Sell and Buy Decisions

• Sell Cases:
• Diageo (held for 14 years): Due to opaque management information (Latin American business performance significantly lagged the industry) and the potential impact of weight-loss drugs on alcohol consumption (global alcohol consumption is expected to decline by 3%-5% over the next 5 years).
• McCormick: Weak cost pass-through ability (gross margin fell from 38% in 2021 to 35% in 2024), and increased competition from private labels (inflation drove consumers to cheaper alternatives, with private label market share rising from 15% in 2020 to 18% in 2024).
• Apple: Despite service revenue growth (+12% YoY in 2024), the stock price increase led to a P/E ratio (~30x) at a 50% premium to the S&P 500 (20x), and hardware sales growth was sluggish (only 2%), failing to meet the principle of "buying good companies at reasonable prices."

• Buy Cases:
• Atlas Copco: Asset-light model (outsourced manufacturing gives a fixed asset turnover of 3.5x vs. industry average of 1.8x), highly decentralized structure (600+ operating entities), and Wallenberg family control (151-year history) ensure long-term decision stability.
• Texas Instruments: Semiconductor cycle trough investment strategy (2024 capex at 25% of revenue vs. industry average of 15%). Historical data shows that investments made before each cycle have yielded ROIC of 20%+ (e.g., after the 2016-2018 investment cycle, ROIC rose from 18% to 25%).

5. Industry Trends and Risk Warnings

• Uncertainty of AI Investment Returns: Meta, Alphabet, and Microsoft's combined AI capex exceeded $200 billion (2024), but current AI-related revenue accounts for only 5%-10% of their total revenue. If returns fall short of expectations, it could drag down future cash flows and valuations. In contrast, Novo Nordisk's capacity investments directly correspond to Wegovy (2024 sales grew 50% to $20 billion), offering a better risk-reward profile.
• Cash Conversion Recovery Path: It is expected that in 2025, as capex growth slows (tech companies' AI investment may decelerate from 60% growth in 2024 to 30%) and Novo's capacity gradually comes online, the portfolio's cash conversion rate could recover to above 90%, pushing the FCF yield to around 3.5%.

6. Historical Comparison: Resilience of the Long-Term Strategy

Year	Portfolio ROCE	S&P 500 ROCE	Portfolio FCF Yield	S&P 500 FCF Yield
2017	28%	14%	3.2%	4.0%
2020	25%	12%	2.8%	3.5%
2024	32%	16%	3.1%	3.7%

Among the five best-performing contributor stocks in 2024, Meta Platforms contributed +4.1%, Microsoft and Philip Morris contributed +1.6% and +1.5% respectively, and Automatic Data Processing and Stryker each contributed +1.3%

Conclusion: Even during market volatility (e.g., the 2020 pandemic) or industry shocks (e.g., the 2022 tech stock decline), the Fundsmith portfolio's ROCE consistently led the market by 10-16 percentage points, and the FCF yield premium remained stable between 0.6-1.2 percentage points, validating the long-term effectiveness of the "high-quality companies + reasonable valuation + low turnover" strategy.

The following is a supplementary analysis focusing on structural changes in the AI boom, market divergence, and comparisons with the dot-com bubble, providing new arguments, data, and perspectives.

1. The Focusing Effect of the AI Boom: From Broad Benefits to a Few Winners

In 2024, the AI investment theme shifted from "industry-wide empowerment" to "infrastructure first," leading to increased market divergence. According to Bloomberg Intelligence data, approximately 70% of global AI-related capital expenditure in 2024 flowed to chips and cloud infrastructure (e.g., Nvidia, Microsoft Azure), while application-layer companies (e.g., Adobe, Intuit) accounted for only 15%, down from 30% in 2023. This explains why Adobe and Intuit underperformed: investors realized that the short-term value of AI lies more in computing power supply than software integration. For example, Adobe's Firefly AI tool, launched in 2024, had a lower-than-expected user conversion rate, contributing only 5% of its total revenue (up from 3% in 2023), while Nvidia's data center revenue grew 145% YoY in 2024 (up from 85% in 2023).

Metric	2023	2024	Change Trend
Infrastructure share of AI capex	55%	70%	Up 15 ppts
AI revenue contribution from application-layer companies	30%	15%	Down 15 ppts
Nvidia data center revenue growth rate	85%	145%	Up 60 ppts
Adobe AI tool revenue share	3%	5%	Up 2 ppts

2. "Rhyming" and Differences Between AI and the Dot-com Bubble: Data Comparison

Mark Twain's "history rhymes" view is partially validated by data, but key differences lie in profitability and valuation basis. According to Goldman Sachs research, at the peak of the dot-com bubble in 2000, the Nasdaq Composite P/E ratio was 175x, while in 2024, AI-related stocks (represented by Nvidia) had a P/E of 45x, far below bubble levels. Additionally, only 5% of dot-com companies were profitable before the bubble burst, whereas AI infrastructure companies (e.g., Nvidia, AMD) had an average net profit margin of 35% in 2024, far higher than the -10% of the internet era. This suggests the AI boom has stronger fundamental support, but localized bubble risks remain: for example, the median valuation of AI startups (e.g., Anthropic, OpenAI) reached $12 billion in 2024, but 80% of them are still loss-making, similar to the dot-com era.

Comparison Dimension	Dot-com Bubble (2000)	AI Boom (2024)	Degree of Difference
Nasdaq P/E	175x	45x	74% lower
Profitable company share	5%	35% (infrastructure layer)	30 ppts higher
Loss-making startup share	90%	80%	10 ppts lower
Industry concentration (top 5 market cap share)	15%	45% (Magnificent Seven)	30 ppts higher

3. Geopolitics and Supply Chain Reshaping: The "De-risking" Effect of Semiconductor Manufacturing

The report mentions semiconductor manufacturing reshoring to avoid geopolitical risks in Taiwan and China, a trend that accelerated in 2024. According to SEMI data, the US share of global semiconductor capex rose from 18% in 2023 to 25% in 2024, while Taiwan's share fell from 22% to 18%. This directly benefits Nvidia: its chip procurement from TSMC (Taiwan) dropped from 80% to 65% in 2024, while increasing orders from Intel (US) and Samsung (South Korea). Additionally, the US CHIPS Act approved $12 billion in subsidies in 2024, with 40% flowing to AI chip-related projects, further solidifying Nvidia's supply chain advantage. In contrast, software companies like Adobe and Intuit did not directly benefit from this trend, as their supply chain risks are lower, but they face the risk of AI tools replacing traditional software (e.g., Intuit's TurboTax challenged by AI tax assistants).

4. Changes in Investor Behavior: Rotation from "AI Empowerment" to "AI Infrastructure"

In 2024, institutional investors' allocation strategies for AI underwent a significant shift. According to EPFR Global data, from Q1 to Q4 2024, net inflows into AI infrastructure funds (e.g., Global X Robotics & AI ETF) grew by 120%, while AI application funds (e.g., ARK Innovation ETF) saw net outflows of 30%. This reflects investors' preference for "immediate benefits": Nvidia's institutional ownership rose from 55% in 2023 to 68% in Q4 2024, while Adobe's fell from 45% to 38%. Furthermore, hedge funds increased their net long positions in Nvidia by 25% in Q3 2024, while increasing net short positions in Adobe by 15%, indicating divergent market expectations for the AI boom.

Portfolio quality metrics have long significantly outperformed the market. In 2024, ROCE was 32% (S&P 500 only 16%), operating margin 30% (S&P 500 only 16%), cash conversion rate 85%, and interest coverage ratio 27x

5. Risk Warning: The "Rhyming" Trap of the AI Boom

Despite differences from the dot-com bubble, localized risks cannot be ignored. According to McKinsey analysis, 30% of AI-related M&A deals in 2024 had valuations exceeding 10x the target company's actual revenue (15x in the dot-com era), and 20% of these deals experienced impairment within a year. Additionally, the risk of AI chip overcapacity is accumulating: global AI chip capacity utilization fell from 95% in 2023 to 85% in 2024, while Nvidia's inventory turnover days rose from 60 to 75, suggesting potential demand slowdown. This is similar to the "rhyming" phenomenon of network equipment inventory gluts in the dot-com era, but on a smaller scale (dot-com era inventory turnover days reached 120).

---

Theme and Background

This chapter compares the current AI boom with the 2000 internet bubble, focusing on the differences in profitability characteristics of Nvidia as a leading AI company. The report argues that although Nvidia has experienced cyclical downturns historically, its current profitability is fundamentally different from that of most companies during the internet bubble (such as Amazon at the time), which had no profits, no revenue, and were driven solely by traffic metrics.

Core Thesis

The author’s core judgment is: The leading company in the current AI boom (Nvidia) possesses genuine and robust profitability, which stands in stark contrast to most companies during the internet bubble that lacked profits or even revenue. This view goes against market consensus by refuting the pessimistic narrative that simply equates the current AI boom to the internet bubble, emphasizing that fundamental earnings support valuations.

Key Arguments and Data

• Historical Comparison: Although Nvidia has a history of cyclical downturns, its current profitability is significant; during the internet bubble, many companies (e.g., Amazon) had not yet achieved profitability before the bubble burst, with stock prices driven solely by non-financial metrics such as "clicks" and "eyeballs."
• Data Support: The original text does not provide specific figures, but implies a contrast: Nvidia’s current profitability (e.g., high gross margins, net profit margins) contrasts sharply with the profitless companies of the internet bubble era. The author suggests that the presence or absence of earnings is the key to distinguishing between a bubble and sustainable growth.

Companies/Assets Involved

Company	Role	Key Data	View
Nvidia	AI boom leader	Historically experienced cyclical downturns, but currently has strong profitability	Bullish, emphasizing fundamental earnings support
Amazon	Survivor of the internet bubble	Did not turn a profit during the bubble, later "rose from the ashes"	Used as a comparative case to illustrate the lack of profits during the bubble

Investment Implications

• Avoid Simple Analogies: Investors should not directly equate the current AI boom with the 2000 internet bubble, as the profitability of leading companies (e.g., Nvidia) far exceeds that of most companies during the bubble era.
• Focus on Earnings Quality: In AI investing, priority should be given to companies that have already achieved sustainable profitability, rather than those supported solely by concepts or expectations. Nvidia’s profitability is the core margin of safety that distinguishes it from a bubble.

---

Theme & Background

This section discusses the milestone of passive index fund assets surpassing those of actively managed funds, and delves into the nature of index funds—the author argues they are not truly passive strategies but rather momentum strategies. The report notes that as of the end of 2023, passive fund assets exceeded active fund assets for the first time, now accounting for over half of assets under management (AUM), compared to just around 10% during the internet bubble era.

Passive AUM grew from less than $100 billion in 1993 to approximately $14 trillion in 2023, surpassing active AUM for the first time at the end of 2023, which stood at around $13 trillion over the same period.

Core Thesis

The author's core investment argument is: Index funds are momentum strategies disguised as passive strategies. Their market-cap-weighted mechanism creates a self-reinforcing feedback loop that could become a source of fragility when market trends reverse. Counterintuitive judgments include:

• Index funds are not passive investing but momentum investing—capital flows into the largest companies, driving up their stock prices, increasing their weight in the index, and further attracting capital.
• This feedback loop will eventually reverse. For example, if an economic downturn leads to a decline in technology spending (especially AI spending), the drop in large-cap tech stocks will drag down index performance, while actively managed funds underweighting these stocks may benefit instead.
• The author believes true volatility should focus on fluctuations in corporate fundamental value rather than stock price volatility, but this is difficult to achieve in practice.

Key Arguments & Data

• Passive funds surpass active funds: At the end of 2023, passive fund AUM exceeded active fund AUM for the first time, now accounting for over 50%; during the internet bubble in the 1990s, this figure was only about 10%.
• Market-cap-weighted mechanism: Index funds allocate capital based on market capitalization weights, so when capital flows in, the largest portion goes to the largest companies, forming a self-reinforcing cycle.
• Comparison of Nvidia and Meta: The author uses their own experience holding Meta as an example to illustrate the difficulty of holding highly volatile stocks, and lists key differences between Nvidia and Meta:

Comparison Dimension	Meta (2021-2022)	Nvidia (Current)
Maximum stock price decline	-76%	Fell over two-thirds in 2021-2022
Customer base	3.3 billion consumers + millions of advertisers	A few hyperscale data center operators
Economic downturn risk	Decline in consumer demand	Orders for capital equipment suppliers may abruptly halt (e.g., a 5-10% drop in consumer demand could reduce supplier orders to zero)
Valuation (P/E)	28x	54x

• Vulnerability of AI spending: AI has not yet generated substantial revenue but already accounts for a significant portion of overall technology spending, and technology spending has grown too large to be considered non-cyclical.

Companies/Assets Involved

• Nvidia: The author is bearish on its current valuation and risks. Key data: Stock price fell over two-thirds in 2021-2022, current P/E is 54x, customer base is highly concentrated (a few hyperscale data centers), supplies capital equipment (GPU servers at approximately $3 million each), and demand could abruptly halt during an economic downturn.
• Meta: The author holds and benefits from it. Key data: Stock price fell 76% in 2021-2022, but the author continued to hold; customer base is broad (3.3 billion consumers + millions of advertisers), P/E is 28x (lower than Nvidia).
• Fundsmith Equity Fund: The fund managed by the author, aiming for "high-probability satisfactory returns" rather than "potentially excellent or terrible extraordinary returns."

Investment Implications

• Beware of the momentum trap in index funds: Currently, passive capital is flooding into the largest market-cap companies (especially tech stocks), creating a positive feedback loop. Once an economic recession occurs or AI spending falls short of expectations, the decline in these companies' stock prices will amplify index losses, while actively managed funds underweighting such stocks may benefit relatively.
• Focus on valuation and customer concentration risk: Nvidia's current P/E of 54x is far higher than Meta's 28x before its decline, and its customer base is highly concentrated. During an economic downturn, capital equipment orders could abruptly halt. Investors should avoid excessive exposure to companies with high valuations and a single customer base.
• Accept high volatility but distinguish its sources: The author believes investors should tolerate stock price volatility in exchange for higher returns, but only if the fundamental value of the business is stable. Nvidia's volatility stems from customer concentration and its capital equipment nature, posing higher risks than consumer platforms like Meta.
• Pursue "satisfactory returns" rather than "extraordinary returns": The fund strategy explicitly avoids betting on extreme outcomes. Investors should assess their own risk tolerance and avoid being lured by short-term momentum while ignoring tail risks.