💫AI Eats the World: Who Captures the Value in the Next AI Platform Shift?

In his latest talk and report, AI Eats the World, Benedict Evans draws a clear distinction between the hype surrounding artificial intelligence and the nature of real technological transformation. In his view, the questions worth revisiting again and again are how generative AI will actually change the world—and who, in the process, will ultimately capture the value and profits.

The report has attracted wide attention not because it names clear winners, but because it places the current AI boom back into a much longer technology cycle: the emergence of a new platform shift.

1. Historical perspective: More like a platform shift than a simple technology upgrade

In Evans’ analytical framework, generative AI is not merely an incremental improvement in software functionality. Instead, it represents a change on the same level as the PC, the internet, and smartphones.

Looking back over the past several decades, every major platform shift has shared a few common characteristics:

• New technological capabilities emerge, redefining what is considered feasible;

• New entry points and distribution channels form, weakening old gatekeepers;

• The overall market expands significantly, yet the earliest participants often fail to capture most of the long-term value.

Evans points out that the biggest difference between this AI wave and previous platform shifts is that we have very little idea where its ceiling lies.

In the PC or mobile eras, performance, cost, and physical constraints were relatively well understood. By contrast, the core capabilities of generative AI come from the interaction of model scale, data, and computing power. We do not fully understand why these models have become so effective, nor do we know where the limits of improvement might be.

This helps explain the sharp divergence in today’s market views:

on one side, bold visions of artificial general intelligence;

on the other, deep skepticism around commercialization paths and valuation.

2. Unprecedented Capex Surge

While the long-term impact of AI remains uncertain, one thing is no longer in doubt: capital has already entered the market at scale.

Across multiple key charts in AI Eats the World, Evans repeatedly emphasizes the same point—AI is triggering an unusually aggressive infrastructure investment cycle.

Large platform companies such as Microsoft, Google, Amazon, and Meta are ramping up investments in data centers, compute clusters, network architecture, and energy infrastructure at an unprecedented pace. In terms of scale, these investments are now comparable to those seen in traditional capital-intensive industries.

$Microsoft(MSFT)$ $Alphabet(GOOG)$ $Alphabet(GOOGL)$ $Amazon.com(AMZN)$ $Meta Platforms, Inc.(META)$

The report cites a growing consensus within the tech industry:

The risk of not investing now is greater than the risk of overinvesting.

This is not a judgment about short-term returns, but about future competitive positioning.

If AI truly becomes a core productivity tool, being absent from the infrastructure layer could mean losing long-term influence and strategic control.

As a result, the current wave of AI development shows a very clear reality: the industry’s center of gravity has shifted first toward the expansion of foundational capabilities, rather than profitability at the application layer.

3. Competition shifts toward products and distribution

One of the most important—and often overlooked—judgments in the report concerns the value of large models themselves.

Evans notes that, based on observable outcomes so far, the performance gap between leading foundation models in general-purpose tasks is narrowing. Across coding, writing, and reasoning benchmarks, results increasingly converge.

This does not mean models no longer matter; rather, it suggests that models are gradually becoming a broadly accessible underlying capability, rather than the sole determinant of competitive advantage.

As models begin to resemble electricity or cloud computing, the focus of competition naturally moves upward. What truly differentiates companies is no longer parameter count, but the ability to turn models into usable products:

• Can they be seamlessly embedded into existing tools?

• Is the learning curve low enough for real adoption?

• Who can turn users from “occasional experimentation” into high-frequency dependence?

Evans highlights a very pragmatic reality: many AI products already have massive user bases, yet actual usage frequency remains low. Many people recognize that AI is powerful, but their way of working has not fundamentally changed.

In other words, widespread awareness does not automatically translate into habit formation—let alone commercial success.

For this reason, the next phase of generative AI competition is likely to resemble traditional software markets more than a model arms race. The key question will not be whose model is stronger, but who truly changes how people get things done.

4. Conclusion: For investors, this is an unfinished structural transformation

Taken together, AI Eats the World offers three conclusions worth revisiting:

• Generative AI resembles a long-term platform shift rather than a short-term thematic trade;

• The most certain changes have already occurred at the capital and infrastructure level;

• As models become infrastructure, lasting value is more likely to emerge from products, distribution, and structural reinvention.

This suggests that, instead of rushing to identify the “next breakout AI application,” investors may be better served by observing: Who is building irreplaceable foundational capabilities—and who is truly turning AI into a tool that people use repeatedly.

🌟Question for you:

• As model capabilities converge, do you see more value in infrastructure providers (compute/cloud), platform companies, or deeply embedded, verticalized application players?

• Do you think current AI valuations reflect long-term structural change, or are they pricing in future usage ahead of reality?

• In this capital expenditure cycle, which segment of the AI value chain are you watching most closely?

Welcome to share your views in the comments.

For SG users only, Welcome to open a CBA today and enjoy access to a trading limit of up to SGD 20,000 with unlimited trading on SG, HK, and US stocks, as well as ETFs.

🎉Cash Boost Account Now Supports 35,000+ Stocks & ETFs – Greater Flexibility Now

Find out more here.

Complete your first Cash Boost Account trade with a trade amount of ≥ SGD1000* to get SGD 688 stock vouchers*! The trade can be executed using any payment type available under the Cash Boost Account: Cash, CPF, SRS, or CDP.

Click to access the activity

Other helpful links:

• 💰Join the TB Contra Telegram Group to Get $10 Trading Vouchers Now🎉

• How to open a CBA. How to link your CDP account. Other FAQs on CBA. Cash Boost Account Website.