全球价值投资协会

Our Perspective | Bubbles and True Gold: Choices in the AI Era from a Value Investment Lens

文章免費5 天前

I. Staying True to Value Amid the AI Frenzy

Since 2026, global AI investment is projected to exceed $700 billion. The computing power investment race has intensified, with concept hype emerging one after another, and the market is permeated with the fear of missing out (FOMO) on AI. Ray Dalio of Bridgewater Associates has warned of a "relatively high level" of bubble in the AI market, with characteristics highly similar to the 2000 dot-com bubble — valuations detached from fundamentals, speculative capital clustering, and paper wealth far exceeding actual cash flows.

As value investors who adhere to the core principles of margin of safety, cash flow supremacy, and long-term economic moats, the Global Value Investment Association (GVIA) firmly believes that technological revolution ≠ investment feast, and conceptual carnival ≠ value realization. AI is not a short-term speculation theme, but a general-purpose technology that will reshape the global economy. The current market is not a holistic bubble, but a structural differentiation — overheating in the infrastructure layer, undervaluation in the application layer, scarcity of genuine growth, and proliferation of pseudo-concepts. Starting from the underlying logic of value investment, this article penetrates the fog of the AI bubble, distinguishes the true value of the industry, and provides a rational decision-making framework for long-term investors.


II. Deconstructing the AI Industry: Value Stratification from "Storytelling" to "Delivering Results"

The AI industry chain presents a clear three-tier structure, with distinct investment logics and risk-return characteristics for each tier. Value investment requires precise stratification and avoidance of the virtual to focus on the real.

(I) Infrastructure Layer: Capital-Intensive "Water Sellers" with High Certainty, Focus on Leading Effect

Core Components: GPU/TPU chips, servers, cloud computing, high-speed networks, liquid cooling, and other computing power infrastructure.

Value Logic: AI training demand is rigid and nearly infinite, and each round of model iteration drives exponential growth in computing power demand, making it a "must-have" track. Global tech giants (Microsoft, Google, Amazon, etc.) have invested over $1 trillion in data centers and semiconductors in three years, with capital expenditure scales comparable to energy giants.

Bubble Risks: Valuations in the infrastructure layer are generally at historical highs in 2026. Some targets have seen "surging profits but stagnant stock prices", with increasing expectations of slowing capital expenditure growth, and marginal changes are likely to trigger valuation corrections.

Value Screening: Prioritize leading companies with stable cash flows, controllable capital expenditures, and high technical barriers, and avoid targets with pure concepts, high debt, and no independent technology.

(II) Model Layer: "Arms Race Zone" of Burning Money, Fast Technological Iteration and Difficult Profitability, Invest Cautiously

Core Components: General large models, industry-specific models, multimodal model R&D enterprises.

Value Logic: High technical barriers, rapid iteration, significant head effect, and may form an oligopoly pattern in the long term.

Bubble Risks: Over 90% of model enterprises are unprofitable, with high daily GPU costs and unclear commercialization paths; there is great uncertainty in technical routes, and today's leaders may be disrupted by new technologies tomorrow.

Value Screening: Focus on leading companies with sustainable R&D investment, high-quality compliant data, and clear scenario implementation, and be wary of enterprises with "bottomless money-burning".

(III) Application Layer: "Golden Zone" of Value Realization, Need Careful Screening and Deliberate Action

Core Components: AI + vertical scenario solutions such as healthcare, finance, industry, education, and office, AI agents, etc.

Value Logic: Directly connect to terminal demand, with clear technology-scenario-business closed loops, which can be quickly converted into revenue and cash flow. The core inflection point of the industry has arrived in 2026 — the investment focus has shifted from "infrastructure frenzy" to "application supremacy", and vertical AI has become the consensus direction.

Market Status: The penetration rate of the application layer is less than 20%, a large number of real demands are unmet, valuations are generally lower than the infrastructure layer, and there are significant value depressions.

Value Screening: Focus on "three-high" enterprises with high order conversion rates, high customer renewal rates, and high AI revenue share. Prioritize high-value-added scenarios such as financial compliance, medical diagnosis, and industrial quality inspection. For projects with existing scenario implementation and broad scenario ceilings, we should have firm investment confidence at reasonable valuations.


III. Distinguishing AI Bubbles: Criteria for Judging "Genuine vs. False" for Value Investors

The core of value investment is to penetrate appearances and return to essence. Faced with the bubble controversy in the AI market, we propose four judgment dimensions to distinguish "genuine growth" from "false bubbles".

(I) Financial Dimension: Cash Flow is the Touchstone

Genuine Value: Positive operating cash flow, stable free cash flow, AI investment supported by internal ,debt/operating cash flow ratio below 3 times, high financial margin of safety.

False Bubble: Continuous losses, reliance on external financing to burn money, revenue cannot cover computing power costs, negative cash flow, high financial vulnerability.

(II) Technical Dimension: Moat is the Dividing Line

Genuine Value: Possess self-developed core technologies, high-quality patents, strong data barriers, and rapid iteration capabilities, which are difficult to replicate or replace.

False Bubble: Technology relies on outsourcing, no core patents, poor data quality, serious homogenization, only relying on concept packaging to attract capital.

(III) Commercial Dimension: Implementation Capability is the Key

Genuine Value: Large-scale implementation of AI business, strong customer willingness to pay, clear business models (subscription, solution fees, etc.), and sustained revenue growth.

False Bubble: No actual implementation scenarios, low customer willingness to pay, vague business models, only relying on "AI+" concept hype to push up stock prices.

(IV) Valuation Dimension: Margin of Safety is the Bottom Line

Genuine Value: Valuation matches performance growth rate, PE is lower than industry average, PEG is less than 1, with long-term holding margin of safety.

False Bubble: Valuation is seriously detached from fundamentals, PE is as high as dozens or even hundreds of times, no performance support, only relying on expectation hype.


IV. Value Investment Strategy: Stay True to Fundamentals While Seizing Opportunities in the AI Wave

(I) Core Principle: Be Fearful When Others Are Greedy, and Greedy When Others Are Fearful

Do Not Chase Hotspots: Stay away from AI concept stocks with short-term surges and no performance support, and reject speculative behaviors of "chasing gains and cutting losses".

Focus on Leaders: Prioritize allocating to industry leaders with stable cash flows, deep moats, and leading technologies, especially enterprises with synergistic advantages of "AI + traditional main business".

Layout Value Depressions: Focus on high-quality targets in vertical fields of the application layer, which currently have low valuations and large growth space, and are the core battlefield for value investment.

Establish a Core Risk Management Framework:

  • Valuation ManagementThe overall valuation of the AI sector is currently high. If liquidity tightens, performance falls short of expectations, or black swan events occur, high-valuation targets may face a 30%-50% correction risk.

  • Technical Risk ManagementAI technology iterates extremely fast. Today's leading technology may be disrupted by new technologies tomorrow, and corporate moats may disappear quickly.

  • Geopolitical RiskIntensifying Sino-US tech competition and restrictions on core technologies such as chips and computing power may affect enterprises' technological R&D and commercialization processes.

  • Commercialization Disappointment RiskThe implementation of AI applications is affected by industry characteristics, customer acceptance, costs and other factors, and the commercialization progress of some scenarios may be lower than market expectations.

(II) Specific Target Selection Principles: Three Priorities and Three Avoidances

Prioritize Allocation to:

  • Computing power infrastructure leaders: GPU, cloud computing, and liquid cooling leaders with stable cash flows and high technical barriers.

  • Vertical application champions: Segment leaders in AI + healthcare, finance, industry and other fields with strong implementation capabilities and good cash flows.

  • Traditional leaders empowered by AI: Enterprises with strong main business cash flow, smooth AI transformation, and reasonable valuations.

Firmly Avoid:

  • Pure concept speculation stocks: "Three-nothing" AI concept stocks with no technology, no implementation, and no cash flow.

  • Model enterprises with bottomless money-burning: Small and medium-sized model companies with continuous losses, reliance on financing, and hopeless commercialization.

  • Overvalued targets with overdrawn performance: Bubble stocks with PE far exceeding industry averages and performance growth rates unable to support valuations.


V. Conclusion: Transcend Bubbles and Embrace True Value

AI is not a fleeting concept hype, but a once-in-a-century technological revolution that will reshape the global economic pattern and create huge value in the long term. However, the short-term market bubble cannot be ignored. Value investors need to maintain rationality and restraint, not be swayed by emotions or confused by concepts, otherwise even the best track will become a Waterloo for investors.

2026 is a watershed year for AI investment — bubbles will gradually clear, and true value will stand out. As value investors, we should adhere to the core logic of margin of safety, cash flow supremacy, and long-term economic moats, avoid bubbles, focus on true gold, and prioritize allocating to computing power leaders with stable cash flows, application champions with strong implementation capabilities, and traditional leaders empowered by AI.

Upholding the way of long-term value investment, GVIA believes that in the AI era, value investment is not an outdated dogma, but the only right way to transcend bubbles and earn long-term stable returns.



Global Value Investment Association (GVIA)June 8, 2026