Former a16z Partner's Major Tech Report: How AI is Eating the World

“AI is eating the world, and we haven't even seen its true form.” In the newly released “AI eats the world” report, renowned tech analyst and former a16z partner Benedict Evans delivers a judgment enough to shake the entire tech world: Generative AI is triggering a once-every-ten-to-fifteen-years platform migration in technology, and we still do not know where it will ultimately lead. Evans points out that from mainframes to PCs, from the internet to smartphones, the foundation of the tech industry is entirely rewritten every decade or so. The emergence of ChatGPT in 2022 is very likely the starting point for the next “fifteen-year transformation.” Global tech giants are rushing into an unprecedented investment race. Microsoft, Amazon AWS, Google, and Meta are projected to reach **$400 billion** in capital expenditure in 2025—a figure surpassing the **annual $300 billion** global telecom investment. “**Underestimating AI is a far greater risk than over-investing,**” says Microsoft CEO Sundar Pichai, as quoted in the report, laying bare the industry's underlying anxiety. The report also cites the 1956 US Congressional automation report and the case of elevator operators disappearing as a reminder: **When technology truly lands, it quietly becomes infrastructure and is no longer called 'AI'.** ## Another Fifteen-Year Transformation: The Historical Law of Platform Migration Evans notes that the tech industry experiences platform migration every ten to fifteen years—from mainframes to PCs, from the web to smartphones—each migration reshaping the entire industry. Microsoft's case exemplifies the brutality: once owning nearly 100% market share in the PC OS era, it became almost irrelevant when the focus shifted to smartphones. Data shows Microsoft's OS share in global computer sales dropped dramatically from its high before 2010, falling below 20% by 2025. Similarly, Apple, which dominated the early PC market, was marginalized by IBM-compatible machines. Evans emphasizes that early leaders often disappear; this seems to be the iron law of platform migration. But three years on, the form of this migration remains unknown. Evans lists early failed ideas from the internet and mobile internet eras, such as AOL, Yahoo portals, Flash plug-ins. Now with generative AI, the possibilities are just as dizzying: browser form, agent form, voice interaction, or some entirely new user interface paradigm—no one truly knows the answer. ## Unprecedented Investment Frenzy: $400 Billion Gamble Tech giants are investing in AI infrastructure at an unprecedented scale. Microsoft, AWS, Google, and Meta are estimated to spend $400 billion in capital expenditure in 2025; by contrast, the global telecom industry's annual investment is about $300 billion. Even more notably, this 2025 growth plan has nearly doubled within the year. US data center construction is overtaking office building construction as the new investment driver. Nvidia faces supply bottlenecks due to inability to keep up with demand, with its quarterly revenue surpassing Intel’s accumulated over years. TSMC is equally unable or unwilling to expand capacity fast enough for Nvidia’s orders. According to Schneider Electric’s industry survey, the main limiting factor for US data center growth is public electricity supply, followed by chip access and fiber optics. US electricity demand is rising about 2%, and AI may add another 1%. While not a problem in China, it's hard to build quickly in the US. ## Model Homogenization: The Moat Disappears, AI May “Commoditize” Despite massive investment, the gap between top large language models in benchmark testing is narrowing to single-digit percentages. Evans warns: > **If model performance converges, large models may become a 'commodity,' and value capture will be reshuffled.** In the most general benchmark tests, the gap between leaders is already very close, with model leadership changing weekly. This suggests models are commoditizing, particularly for general use. Evans notes that after three years of development, there's more scientific and engineering progress, but still little clarity on market form. While models continue to improve—more models, Chinese vendors, open source projects, new technical acronyms—clear moats have not emerged. In his view, AI companies must find new moats in computing scale, vertical data, product experience, or distribution channels. ## User Engagement Dilemma: ChatGPT’s 800 Million Weekly Actives Cannot Hide Real Stickiness Shortfall Although ChatGPT claims 800 million weekly active users, user engagement data paints a different picture. Multiple surveys show only about 10% of US users use AI chatbots daily; most are still merely dabbling. Deloitte’s survey shows far more people occasionally use AI chatbots than those who use them daily. Evans calls this a classic “**engagement illusion**”: AI’s adoption rate is astonishing, but it is not yet an everyday tool at the mass level. He analyzes the reasons for the engagement dilemma: How many use cases are obviously and easily adaptable? Who has a flexible work environment and consciously seeks optimization? For everyone else, does AI need to be packaged into tools and products? This highlights a significant gap between technical capability and practical application. Enterprise deployment is equally slow. The report cites surveys by consulting agencies showing that although enthusiasm for AI is high, few projects have actually entered production environments. - **Deployed: 25%** - **Planned for H2 2025: around 30%** - **Earliest deployment in 2026: around 40%** Currently, success cases are mainly in coding assistance, marketing optimization, automatic customer support—still in the “**absorption stage**,” a long way from true business restructuring. ## Advertising and Recommendation Systems Enter Disruptive Rewrite Evans believes that advertising and recommendation systems are the fastest areas to be radically transformed by AI. Traditional recommendation relies on “relevance,” whereas AI can understand “user intent” itself. This means: > **The underlying mechanisms of the trillion-dollar advertising market may be rewritten.** Google and Meta have already revealed early data: AI-driven ads can deliver a **3%–14% uplift in conversion rates**. The cost of creative ad production, which makes up a $100 billion annual market, could be further reshaped by automated generation technology. ## Historical Lesson: When Automation Succeeds, It Is No Longer Called ‘AI’ Evans draws back to the 1956 US Congressional automation report, noting that each wave of automation sparks heated social debate but ultimately quietly becomes infrastructure. The disappearance of elevator operators, the barcode-driven inventory revolution, the internet’s evolution from “novelty” to infrastructure—all prove: > **When technology truly lands and becomes universal, people no longer call it ‘AI’.** Evans emphasizes that AI’s future is both clear and ambiguous: We know it will reshape industries, but not the final product form; know it will be everywhere in enterprises, but not who will dominate the value chain; know it will need massive compute, but not where growth will stop. Put another way, AI is becoming the protagonist of the new fifteen-year cycle, but the script of the drama is still unwritten. > **We may be standing on the fault line of the next tech earthquake.** ## The Future of Value Capture: From Network Effects to Capital Competition For research- and capital-intensive commoditized products, value capture becomes crucial. If models become commodities and lack network effects, how will model labs compete? Evans suggests three possible paths: expanding downstream for scale, expanding upstream through network effects and product, or identifying new competitive dimensions. Microsoft’s case demonstrates a shift from network-effect-based competition to capital-access-based competition. The company’s capital expenditure as a percentage of sales revenue has risen sharply from historic lows, reflecting a fundamental change in competition dynamics. OpenAI, meanwhile, has adopted an “always say yes” strategy, including infrastructure deals with Oracle, Nvidia, Intel, Broadcom, AMD; e-commerce integration, advertising, vertical data sets, and diversified layouts involving application platforms, social video, and web browsers. Risk warning and disclaimer Markets carry risks; investment needs to be cautious. This article does not constitute personal investment advice, nor does it take into account individual users’ special investment objectives, financial situations, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific situation. Invest accordingly at your own risk.