Big Tech’s latest earnings season has delivered a familiar kind of reassurance: the numbers are bigger, the margins are holding up, and the businesses that once looked like they were being disrupted by the next wave of technology are now proving—quarter after quarter—that they can still convert innovation into cash. Meta and Alphabet, along with other large platform and cloud-adjacent players, have reported results that investors can point to with confidence. Revenue growth is not just surviving; in several cases it is accelerating. Costs are being managed with discipline rather than desperation. And the core engines—advertising for the social giants, search and cloud for the search-and-platform conglomerates—are still functioning as reliable profit machines.
Yet beneath the surface, the market’s conversation has shifted. The question is no longer simply whether these companies are executing well today. It is whether their execution today is building the kind of advantage that will matter most tomorrow—an advantage that may be defined less by incremental product improvements and more by something harder to measure: AI supremacy.
That phrase sounds dramatic, but it captures a very practical investor anxiety. “Supremacy” is not a single metric. It is not a line item on an income statement. It is not even a consistent set of KPIs across companies. For one firm, it might mean the ability to build frontier models and deploy them at scale. For another, it might mean owning the distribution layer—turning AI into a default interface for billions of users. For a third, it might mean controlling the infrastructure stack: chips, data pipelines, training efficiency, and the operational muscle required to run AI workloads reliably and cheaply.
The problem is that earnings can look strong while the competitive edge required for long-term dominance is still forming. In other words, the market is increasingly asked to value a future that is not yet fully visible, using evidence that is necessarily backward-looking. That tension is what makes this earnings season feel different, even when the results themselves are not.
To understand why, it helps to separate two kinds of performance. There is performance that is measurable immediately—growth in revenue, improvement in operating margin, cash flow generation, and the ability to fund capex without breaking the balance sheet. Then there is performance that is measurable only after a lag—whether AI capabilities translate into durable user engagement, whether monetization scales beyond early pilots, whether developers adopt platforms at a rate that compounds, and whether the company’s model of distribution becomes a moat rather than a feature.
This is where the “ever bigger, ever less useful” dynamic comes from. Earnings are getting larger, but the explanatory power of those earnings is diminishing relative to the size of the question investors are trying to answer. The market can see the past clearly. It struggles to see the future with the same clarity.
Meta’s case illustrates the shift. The company’s results have continued to reflect a business that is not merely stable but improving—advertising demand, engagement patterns, and the company’s ability to monetize attention remain intact. But the market’s focus is increasingly on how AI changes the mechanics of that monetization. If AI improves targeting, creative generation, and ad measurement, then it can raise the value of each impression. If AI improves the user experience—recommendations, content ranking, and moderation—then it can increase time spent and reduce churn. If AI becomes a new interface for discovery, then it could reshape how users find content and how advertisers reach them.
The difficulty is that none of these outcomes are guaranteed, and none of them show up neatly in a single quarter’s earnings report. A company can spend heavily on AI infrastructure and still deliver strong near-term results because the existing advertising machine is still working. But investors want to know whether the AI spending is building a structural advantage or simply buying parity with competitors. They also want to know whether the company’s AI strategy will create a new layer of monetization that is resilient even if traditional ad formats face pressure.
Alphabet faces a similar challenge, though the contours differ. Search remains a core profit engine, and cloud continues to be a major growth and margin story. But AI is changing the definition of “search.” The user intent that used to be satisfied by ranking links is increasingly satisfied by generating answers, summarizing information, and orchestrating multi-step tasks. That means the economics of attention may change. It also means the distribution of queries—what users ask, how they ask it, and where they get answers—could shift away from classic search results pages toward conversational interfaces and agentic workflows.
Alphabet’s earnings can therefore be strong while the market worries about whether the company’s AI approach will preserve its role in the value chain. If AI reduces the number of clicks or changes the way ads are integrated into the experience, then the monetization model could evolve in ways that are difficult to forecast. If, however, AI increases query volume, improves relevance, and strengthens advertiser ROI, then the monetization story could improve faster than expected. The earnings report tells you which direction the company is moving today. It does not fully reveal the shape of the future.
This is the broader pattern across Big Tech: the fundamentals are being proven, but the strategic stakes are being reframed. Investors are not ignoring earnings; they are treating them as necessary but insufficient evidence. The market is increasingly asking: are these companies building the kind of AI advantage that will compound over time, or are they simply participating in a race whose outcome is uncertain?
One reason the question feels so hard to answer is that AI supremacy is not a single contest. It is a portfolio of capabilities that must align. A company can have strong models but weak distribution. It can have distribution but lack the data advantage needed to improve performance. It can have infrastructure but struggle to translate AI into products that users actually prefer. It can have products but fail to monetize them at scale. Supremacy, in practice, is often less about raw technical brilliance and more about the ability to integrate AI into real-world systems—at speed, at reliability, and at cost.
That integration is where earnings become “less useful.” Financial statements are excellent at capturing what has already been built and sold. They are weaker at capturing what is being built but not yet monetized. AI development is particularly prone to this mismatch because it involves long cycles: training runs, evaluation, iteration, deployment, and then continuous improvement based on user feedback. Even when a company is making progress, the payoff may arrive later than the market expects—or arrive in a form that is not easily comparable to prior quarters.
There is also a second mismatch: the market’s valuation framework is struggling to keep pace with the nature of AI competition. Traditional tech valuation often relied on relatively stable assumptions about growth rates, margins, and competitive dynamics. AI introduces uncertainty about cost curves and product substitution. If AI reduces the cost of producing content, customer support, or software development, then it can compress margins across industries. If AI increases productivity and creates new demand, then it can expand markets. Either way, the path from AI capability to economic value is not linear.
That is why investors are increasingly focused on “hard-to-answer questions.” These questions are not just about whether a company is doing AI. They are about whether the company is doing the right AI, in the right way, with the right feedback loops.
Consider the feedback loop problem. AI systems improve when they can learn from data and user interactions. But the best data is often tied to distribution. The best distribution is often tied to user trust and habit. Trust and habit are built over time. So the companies with the strongest user ecosystems may have an advantage that is not captured by near-term earnings. Their AI models can be trained and refined using signals that competitors cannot access at the same scale. Meanwhile, competitors may be able to build impressive models but struggle to deploy them in ways that generate the same quality of feedback.
Now consider the cost curve problem. AI compute is expensive, and efficiency matters. Companies that can optimize training and inference—through better hardware utilization, model compression, caching strategies, and smarter routing—can deliver AI features at lower marginal cost. Lower marginal cost can enable broader deployment, which can generate more data, which can improve models further. This virtuous cycle is powerful, but it is difficult to observe early. Earnings can show capex and operating expenses, but they do not directly reveal whether a company is winning the efficiency race.
Then there is the monetization problem. AI can improve user experience without immediately improving revenue. It can also improve revenue without improving user experience. The market wants both, but it cannot always tell which is happening until adoption reaches a threshold. For example, AI-generated content might increase engagement, but it could also reduce the diversity of content or increase the risk of low-quality outputs. That could affect long-term retention and brand trust. Similarly, AI-driven ad targeting might improve short-term ROI, but if it changes the user journey too much, it could alter the long-term relationship between platforms and advertisers.
These are the kinds of questions that make earnings feel less useful. Not because earnings are wrong, but because they are not designed to answer questions about future competitive structure.
Still, it would be a mistake to treat this as purely pessimistic. The fact that earnings are growing matters. It suggests that these companies are not merely spending money—they are still extracting value from their existing businesses while investing in AI. That combination is rare. Many firms in tech history have had to choose between growth and profitability. Big Tech appears to be doing both, at least for now.
But the market is now asking whether this dual achievement is sustainable. AI investment is not a one-time expense; it is an ongoing commitment. If AI becomes central to product experiences, then the cost base will rise. The key question becomes whether the incremental revenue generated by AI offsets the incremental costs. If it does, then earnings growth can continue and the market’s confidence can deepen. If it does not, then earnings may still grow for a while due to momentum in existing revenue streams, but the quality of that growth could deteriorate.
This is where the “ever bigger” part of the