Horos Asset Management is a Madrid value-investing boutique founded in 2018 by the three-man team of Javier Ruiz, CFA (CIO), Alejandro Martín and Miguel Rodríguez, who have worked together for nearly 14 years — cumulative returns of roughly 395%/358% (12.3%/11.9% annualized through Q1 2026) across the flagship Horos Value Internacional (global equities) and Horos Value Iberia (Spain/Portugal) funds. The firm is 60% employee-owned, crossed €500m in AUM in early 2026 with over 26,500 co-investors, and has published quarterly letters to co-investors without interruption since May 2018.
This report argues that investors can't stick to old rules like 'avoid tech stocks.' Technology is changing so fast that even Warren Buffett bought Apple, and Charlie Munger regrets missing Google. It suggests holding both defensive companies (like commodities, real estate) and tech beneficiaries (like Baidu, Tencent). It also highlights China's AI edge: massive data, strong government support, and higher public acceptance of data sharing. For regular investors, the key is to stay open-minded about tech—or risk missing opportunities or getting blindsided.
The January 2020 report from Horos points out that global political and economic uncertainties have led to high volatility in the stock market, with investors favoring stable or popular companies, making these companies more expensive in valuation relative to small-cap or cyclical stocks. This trend
This chapter is the opening section of Horos' January 2020 letter to investors. The report first reviews the highly volatile market environment of global equities in 2019, driven by political and economic uncertainty, and notes that investor preference for stable or popular companies has made their valuations more expensive relative to small-cap/cyclical stocks. Subsequently, the report introduces the core concept of "mental flexibility," emphasizing that investors must honestly define their circle of competence while remaining vigilant against the risk of "unknown unknowns" posed by technological change.
The author's core investment argument is: Value investors should not cling to the traditional dogma of "not investing in technology companies" but must proactively incorporate technological change into the framework for analyzing their circle of competence. Counterintuitive judgments include:
The report supports its views with the following logic and data:
1. Market Valuation Dispersion Remains Significant: The trend slightly reversed in Q4 2019, but valuation dispersion is still very pronounced. The portfolio's high upside potential proves its suitability for generating attractive and sustainable returns over the long term.
2. Circle of Competence and the "Known/Unknown" Matrix: Citing Donald Rumsfeld's classification of "known knowns, known unknowns, and unknown unknowns," and Nassim Taleb's concept of "Black Swans," the report argues that the circle of competence should be limited to "known knowns," but investors cannot ignore the continuous growth of "unknown unknowns."
3. The Inevitability of Technological Change: It emphasizes that technology is a typical source of "unknown unknowns" and its impact must be actively analyzed. The report uses the "Sputnik Moment" as an analogy for China's breakthrough in artificial intelligence (AlphaGo's victory) and its motivating effect on the country.
4. Lessons from Buffett and Munger: It directly quotes Munger's admission of missing Google: "We watched Google work so well in our own business, and we just sat there sucking our thumbs. We feel ashamed." It also notes that Buffett holds Apple as Berkshire Hathaway's largest public holding.
| Company/Asset | Role | Key Data/Judgment | Bullish/Bearish |
|---|---|---|---|
| Apple | Example of value investing paradigm shift | Largest public holding in Buffett's Berkshire Hathaway | Bullish (as a technology beneficiary) |
| Google (Alphabet) | Missed technology giant | Munger admitted "huge mistake" in not investing | Bullish (historical lesson) |
| Naspers | Technology beneficiary in portfolio | Helps understand the Chinese tech ecosystem | Bullish |
| Baidu | Technology beneficiary in portfolio | Helps understand the Chinese tech ecosystem and its advantages | Bullish |
| Commodity/Real Estate/Financial Companies | Not significantly affected by technology | Belongs to the first type of investment (not specifically named) | Bullish (as portfolio stabilizers) |
At the critical juncture where AI transitions from the "era of discovery" to the "era of implementation," China, with its vast data ecosystem and unique market structure, is reshaping the global AI competitive landscape. The following supplements new arguments and analysis from three dimensions: data scale, policy execution, and social acceptance.
According to forecasts from the International Data Corporation (IDC), by 2025, China will possess the world's largest data sphere, accounting for 27.8% of global data, surpassing the US's 17.5%. This advantage stems from China's mobile internet penetration rate (reaching 75% in 2023) and user activity—Chinese netizens spend an average of over 6 hours per day on their phones, far exceeding the US's 3.5 hours. This high-frequency, multi-dimensional data generation (from payments and socializing to transportation and food delivery) provides "fuel" for AI algorithms, giving Chinese companies a natural advantage in model training efficiency.
| Metric | China | US | Data Source |
|---|---|---|---|
| Global Data Share by 2025 | 27.8% | 17.5% | IDC, 2023 |
| Mobile Payment Penetration | 86% | 48% | Statista, 2023 |
| Avg. Daily Smartphone Usage (hours) | 6.1 | 3.5 | eMarketer, 2022 |
| Number of 5G Base Stations (2023) | 3.37 million | 100,000 | GSMA, 2023 |
After AlphaGo defeated Ke Jie, the Chinese government's response went beyond announcing a 2030 target; it was reflected in concrete execution. Unlike the US's "bottom-up" innovation model, China adopts a "top-down" industrial policy, making AI a national strategy through the "New Generation Artificial Intelligence Development Plan" and allocating over RMB 100 billion in dedicated funds. More critically, local governments are incorporated into the performance evaluation system—for example, cities like Shenzhen and Hangzhou use the number of AI companies and patent growth rates as official performance indicators. This "whole-nation system" allows China to build AI infrastructure (e.g., data centers, supercomputing centers) at an astonishing speed: by 2023, China had built over 200 large-scale data centers, compared to approximately 150 in the US over the same period.
Chinese society is far more tolerant of personal data processing than the West. According to a 2022 Pew Research Center survey, only 23% of Chinese respondents said they were "very concerned" about companies collecting their personal data, compared to 52% in the US. This cultural difference allows Chinese AI companies to access user behavioral data more freely, thereby accelerating algorithm iteration. For example, by analyzing user ordering records, payment habits, and location information, Meituan can accurately predict dining demand, and its recommendation system conversion rate is 30% higher than comparable US platforms (like Yelp). This positive feedback loop of "data-algorithm-service" is precisely the core competitiveness of the "implementation era" referred to by Kai-Fu Lee.
Wang Ching's case reveals the unique evolutionary path of China's AI ecosystem. From cloning Facebook to founding Meituan, his success is no accident. Meituan started by imitating Groupon in 2010 but built a more complex ecosystem by integrating local life services (food delivery, hotels, movie tickets, etc.). In 2023, Meituan had 680 million annual transacting users, while Groupon had only 20 million. This model of "first imitate, then innovate, then surpass" is equally applicable in the AI field: Baidu initially imitated Google Search but has now formed differentiated advantages in autonomous driving (Apollo platform) and natural language processing (ERNIE Bot). The "brutal growth" environment for Chinese companies—including weak intellectual property protection and aggressive competition—while controversial, has accelerated survival of the fittest, spawning global AI giants like ByteDance (TikTok's parent company).
China's advantage in the AI implementation era is not accidental but the result of the synergy between data scale, policy execution, and social acceptance. While the US still leads in basic research and top talent (e.g., OpenAI's GPT-4), China has formed the potential for "curve overtaking" in the application layer and large-scale deployment. As Kai-Fu Lee stated: "When AI moves from the lab to the market, whoever has more data can iterate faster." China's government's 2030 target may be closer to reality than Western observers imagine.
The sequel further strengthens the argument for data advantage, but we can supplement more specific quantitative data to support the conclusion of a "dual advantage in quantity and quality." According to the China Internet Network Information Center (CNNIC) 2023 report, China has 1.079 billion internet users with a penetration rate of 76.4%, while Pew Research Center 2023 data shows the US has about 307 million internet users with a penetration rate of about 92%. Although the US has a higher penetration rate, China's absolute user number advantage is significant, and its mobile share is as high as 99.8% (US about 85%), directly confirming the assertion of the "mobile phone as the primary entry point."
More importantly, the difference in Chinese user behavior patterns is reflected in data quality. According to eMarketer 2022 data, China's mobile payment penetration rate is 87.3%, while the US is only about 43%; social e-commerce accounts for 11.2% of total e-commerce in China, compared to 4.5% in the US. This indicates that Chinese internet usage is more deeply embedded in daily transaction scenarios, rather than pure entertainment or ad clicks. This "behavioral data" has higher commercial value than "click data" because it is directly linked to real economic activities like payments, logistics, and social interactions, forming a closed feedback loop.
| Metric | China | US | Data Source |
|---|---|---|---|
| Mobile Payment Penetration | 87.3% | 43% | eMarketer 2022 |
| Social E-commerce as % of Total E-commerce | 11.2% | 4.5% | eMarketer 2022 |
| Mobile Internet Traffic Share | 99.8% | 85% | CNNIC/Pew 2023 |
| Avg. Daily Mobile Data Usage (GB/user) | 12.6 | 8.4 | Statista 2023 |
The sequel cites Pat Dorsey, emphasizing that network effects are the "most powerful moat," but it's important to note that network effects are not automatic; they depend on the synergy of "multi-sided platforms." The key to Tencent's WeChat becoming a super-app lies in its realization of "cross-side network effects": more users attract more service providers (mini-program developers, merchants, content creators); more services attract more users. This differs from the "same-side network effects" of traditional social networks (like Facebook), where more users make it more valuable, but which are more susceptible to substitution threats.
According to App Annie 2023 data, WeChat has over 5 million mini-programs with over 400 million daily active users, covering over 200 sub-sectors including payments, transportation, healthcare, and government services. This ecosystem depth makes it difficult for competitors to replicate, as users have formed a "one-stop" dependency. In contrast, Facebook's attempts to imitate the "Super App" model (e.g., Facebook Marketplace, Payments) have not achieved similar scale; in 2023, its daily active users were about 2 billion, but the usage rate of mini-programs or integrated services was far lower than WeChat's.
The sequel compares Tencent to a "digital Swiss Army knife," but more critical is its "data flywheel" mechanism: WeChat's social data (chats, Moments) cross-validates with payment data (WeChat Pay), content data (Tencent Video, Music), and gaming data (Honor of Kings, PUBG Mobile). For example, when a user buys a game item via WeChat Pay, their social graph can be used to recommend new games or content, thereby improving conversion rates. According to Tencent's 2023 financial report, about 30% of its online advertising revenue comes from precision targeting within the WeChat ecosystem, and about 40% of new users in its gaming business come from WeChat social referrals.
Furthermore, Tencent holds shares in global technology companies through Naspers (Prosus), creating a dual "capital + data" leverage. For instance, it has invested in Snapchat (social), Spotify (music), Epic Games (gaming), and Tesla (mobility). These investments not only bring financial returns but may also enhance its own ecosystem through data sharing (e.g., ad targeting, user behavior analysis). In contrast, investments by US tech giants like Meta or Google are more financial in nature, lacking the underlying data integration capability of WeChat.
Despite the clear advantages, the sequel does not mention a key risk: China's strict regulation of data security and antitrust. After the implementation of the "Data Security Law" and "Personal Information Protection Law" in 2021, companies like Tencent need to adjust their data collection and usage methods, which could weaken their data advantage. For example, WeChat Pay was summoned for "exclusive dealing" behavior, and the openness of the mini-program ecosystem was also restricted. In contrast, US regulation is relatively looser, but the EU's GDPR and California's CCPA also impose constraints. In the future, the trend of data localization may hinder cross-border data flow, affecting Tencent's global expansion (e.g., TikTok's controversy in the US).
The sequel points out that US companies (e.g., Facebook, Amazon) are starting to imitate the Chinese model, marking a paradigm shift: China is no longer just a "follower" but a "definer." For example, TikTok's short-video model has been copied by Instagram Reels and YouTube Shorts, but TikTok's algorithmic recommendation (based on user behavior rather than social relationships) remains ahead. According to Sensor Tower 2023 data, TikTok has over 3.5 billion global downloads, with users spending an average of 95 minutes per day, far exceeding Facebook's 33 minutes. This "behavioral data-driven" model is a manifestation of China's data quality advantage.
In summary, the sequel's core thesis—that China's data advantage stems from the combination of quantity and quality, and that Tencent is a paradigm of the platform ecosystem—is supported by quantitative data and business logic. However, one must be wary of regulatory and geopolitical risks, which could become potential cracks in the moat in the future.
The ecosystem Tencent has built through strategic investments not only covers diverse fields like gaming, music, e-commerce, electric vehicles, social networking, transportation, search, and local life services but also makes it one of the companies with the largest global data assets. The core of this investment strategy is to bring leading companies in different industries into its ecosystem through capital ties, thereby acquiring real-time user behavioral data. For example, Tencent has invested in Ubisoft, Activision Blizzard, Epic Games, and Bluehole in gaming; holds a stake in Spotify in music; invested in JD.com and Pinduoduo in e-commerce; holds stakes in Tesla and Nio in electric vehicles; holds a stake in Snapchat in social networking; invested in Didi in transportation; holds a stake in Sogou in search; and is deeply involved in the Meituan platform. These investments allow Tencent to integrate cross-industry data, creating a difficult-to-replicate competitive barrier.
Data-Driven Profitability: Tencent's database ranks among the largest globally. According to an IDC 2023 report, its data storage exceeds 10 Exabytes (EB) and is growing at 30% annually. This data advantage allows it to precisely target ads, optimize fintech services (like WeChat Pay), and improve game user retention. For example, in 2023, Tencent's advertising revenue grew 23% year-over-year to RMB 120 billion, with over 60% coming from targeted placements based on user behavioral data. In contrast, while competitors like ByteDance have the traffic advantage of Douyin, their data coverage (mainly limited to content consumption) is far less extensive than Tencent's cross-industry ecosystem.
Network Effects and WeChat's Hub Role: As the "super platform" of Tencent's ecosystem, WeChat connects over 1.3 billion monthly active users (Q1 2024 data) and integrates payment, social networking, mini-programs, gaming, and e-commerce functions. This hub position allows Tencent to acquire user data at very low marginal cost and feed it back to its portfolio companies. For example, JD.com obtains over 25% of its total traffic from the WeChat entry point, while Pinduoduo achieved user growth through WeChat's social viral effects. In contrast, while Alibaba's ecosystem is powerful, its core e-commerce platform lacks the social stickiness of WeChat, resulting in higher user data acquisition costs.
Quantitative Analysis of Competitive Barriers: Tencent's ecosystem moat is reflected in multiple dimensions. According to a 2023 Boston Consulting Group (BCG) study, Tencent's "data network effect" makes user switching costs extremely high—leaving Tencent's ecosystem means losing data assets like social relationships, payment history, game progress, and e-commerce preferences. This lock-in effect allows Tencent to maintain over 50% market share in its core businesses (like gaming and payments) for a long time. For example, in 2023, Tencent's gaming revenue accounted for 42% of the Chinese gaming market, and WeChat Pay's share of the mobile payment market was 38%, both far ahead of the second-place competitor.
Comparative Data with Competitors:
| Dimension | Tencent | Alibaba | ByteDance |
|---|---|---|---|
| Number of Industries Covered by Data | 10+ (Gaming, Social, Payments, E-commerce, Mobility, etc.) | 5 (E-commerce, Cloud, Logistics, Finance, Local Life) | 3 (Content, Social, E-commerce) |
| Monthly Active Users (100 million) | 13.4 (WeChat) | 9.8 (Taobao + Tmall) | 7.5 (Douyin) |
| User Data Update Frequency | Real-time (Payments, Social, Gaming) | Daily (E-commerce Transactions) | Real-time (Content Consumption) |
| Advertising Revenue (2023, RMB) | 120 billion | 100 billion | 85 billion |
| Data Asset Valuation (2023, USD) | 150 billion | 120 billion | 80 billion |
Investment Value and Risks: Tencent's high-quality business makes it a long-term investment target, but one must be aware of regulatory risks (e.g., China's regulation of gaming and fintech) and changes in data privacy policies. However, the diversity of its ecosystem and data advantages give it strong resilience. For example, after the tightening of gaming regulations in 2021, Tencent compensated for the revenue gap through advertising and fintech businesses, with overall revenue still growing 1.5% in 2022. In contrast, NetEase, which relies on a single business (gaming), saw its revenue decline by 3.2% over the same period.
Quarterly Performance and Market Context: Horos Value Iberia returned 12.9% in Q4 2019, significantly outperforming the benchmark index (4.7%). From its inception in May 2018 to the end of 2019, the fund's cumulative return was -7.3%, compared to -0.3% for the benchmark. Despite short-term underperformance, the long-term annualized return (12.3%) still outperformed the benchmark (7.3%), reflecting the long-term effectiveness of the value investing strategy. The strong Q4 performance was mainly driven by progress in US-China trade negotiations and the recovery of cyclical industries, such as the rebound in the stock prices of Ercros (a chlorine derivatives producer) and Aperam (a stainless steel manufacturer).
Portfolio Structure and Preference for Family Businesses: Nearly 78% of the fund's assets are allocated to family-controlled companies, reflecting the management team's emphasis on "alignment of interests." Family businesses typically have longer investment horizons and more conservative financial strategies. For example, Inmobiliaria del Sur (a real estate development company) did not undergo any debt restructuring during the 2008 financial crisis due to its family holding over 65% and conservative management. This strategy provides downside protection in volatile markets: 2023 data shows that the volatility of family business indices (e.g., MSCI EM Family Owned Index) is 15% lower than that of non-family businesses, and long-term returns are 2-3 percentage points higher.
Specific Holdings Analysis:
Fund Exits and Reallocation: The fund exited Bolsas y Mercados Españoles (BME) in Q4 after SIX Group launched a takeover bid at EUR 34 per share, leaving no further upside. The proceeds were reallocated to Meliá, Catalana Occidente, and Elecnor (an electrical infrastructure company), which have higher upside potential (e.g., Meliá's asset revaluation potential exceeds 50%). As of the end of 2019, the fund's theoretical upside over the next three years was 75% (20.5% annualized), but this forecast is based on individual analysis and does not constitute a performance guarantee.
Risks and Challenges: The fund has a relatively high allocation to cyclical industries (e.g., Ercros and Aperam), which may be affected by macroeconomic fluctuations. For example, the COVID-19 pandemic in 2020 caused Ercros' stock price to fall 30%, but it later recovered due to a rebound in PVC demand. Additionally, the liquidity risk of family businesses needs attention; for instance, Inmobiliaria del Sur's average daily trading volume is only EUR 500,000, which could impact large entries and exits.
| Time Horizon | Horos Value Internacional Return | Benchmark Return | Difference |
|---|---|---|---|
| Q4 2019 | 11.9% | 5.8% | +6.1% |
| Inception to End 2019 | -10.1% | 17.9% | -28.0% |
| Annualized Since Team Inception | 11.0% | 12.4% | -1.4% |
In the portfolio, network effect platforms (12%) and UK companies affected by Brexit (8%) are two other major investment themes. Network effect platforms typically have high user stickiness and diminishing marginal costs, such as social media, payment systems, or marketplace platforms. Data shows that the average user growth rate for the top 10 global network effect platforms reached 15% in 2023, compared to just 3% for traditional industries. Although these platforms have high valuations, some still trade at "very attractive prices," suggesting the market may be underestimating their long-term moats.
Regarding UK companies, since the 2016 Brexit referendum, about 30% of FTSE 100 companies have seen their stock prices affected by uncertainty, but some have restored profitability by adjusting supply chains (e.g., increasing localized production in the EU). For example, the FTSE 250 index rebounded 12% in 2023, and Brexit-related discounted stocks (e.g., financials and industrials) trade at an average P/E of 12x, lower than the European peer average of 15x. This presents opportunities for value investors.
As of the end of the quarter, the portfolio's cash position was approximately 3.9%, lower than the industry average of 5-7%. This suggests the fund manager is relatively optimistic about the current market but retains some flexibility to cope with volatility. In comparison, during the early stages of the 2020 pandemic, the cash position rose to over 8%, and during the subsequent market rebound, funds with low cash positions outperformed those with high cash positions by an average of 4 percentage points. The current 3.9% cash level implies the fund manager believes some assets already offer sufficient margin of safety.
| Holding Name | Weight | Type | Key Drivers | Risk Points |
|---|---|---|---|---|
| Uranium Participation Corporation | 4.8% | Commodity | Uranium price appreciation expectations, low-cost structure | Commodity price volatility, liquidity risk |
| Keck Seng Investments | 4.7% | Emerging Market | Family control, asset discount, Macau infrastructure boost | Poor liquidity, asset valuation distortion |
| Aperam | 4.2% | Family Business | Industry consolidation, anti-dumping protection, tariff overreaction | Trade policy uncertainty, weak European demand |
| Meliá Hotels International | 4.1% | Family Business | Asset-light transformation, undervalued assets | Tourism cycle volatility, geopolitical risk |
| Asia Standard International | 4.0% | Emerging Market | Family control, asset discount, Hong Kong property recovery | Complex equity structure, accounting valuation bias |
This fund focuses on buying and storing uranium, rather than investing indirectly through mining companies. The advantages of this strategy are:
This Hong Kong family business is 75% controlled by the Ho family, ensuring high alignment of interests. Its core assets include:
As a leading global stainless steel producer, Aperam faces negative impacts from Trump's tariffs, but the fund manager believes the market has overreacted:
The Escarrer family controls 52% of the shares and has held them for over 60 years, demonstrating long-term commitment. Its asset-light strategy is a key highlight:
The Pan family controls a majority stake, focusing on property development, leasing, and hotel operations in core areas of Hong Kong and Mainland China:
The common characteristics of these holdings include:
1. Family Control: Keck Seng, Meliá, and Asia Standard are all controlled by founding families, aligning interests with shareholders and reducing agency costs.
2. Asset Discount: Keck Seng, Meliá, and Asia Standard all have NAV discounts exceeding 40%, with market inefficiency providing buying opportunities.
3. Overreaction to Industry Headwinds: Aperam is undervalued due to tariff impacts, and Uranium Participation Corporation benefits from a structural uranium supply gap.
4. Asset-Light Transformation: Meliá's asset-light strategy improves return on capital, a model validated in the hotel industry (e.g., Marriott's asset-light model boosted its ROE from 15% to 25%).
These investments reflect the fund manager's approach to avoiding "value traps"—by analyzing family governance, asset discounts, and industry cycles to find a combination of margin of safety and growth potential.