In 2025, enthusiasm about artificial intelligence drove technology stocks to gains that felt, in retrospect, more like momentum than valuation. The sector delivered 26% earnings growth — strong by any historical standard — but valuations stretched well above what that growth justified, with some AI-adjacent names trading at 40–80x forward earnings.
In 2026, the underlying earnings delivery has accelerated: tech sector EPS growth is projected at 43%, according to FactSet consensus estimates, making it the strongest-growing major sector in the S&P 500 by a significant margin. But the market's reaction has been more measured than in 2025, and professional investors are describing their approach with phrases like "pick-your-spots" rather than the broad buying that characterized the first wave.
The divergence between accelerating fundamentals and more cautious sentiment deserves careful examination.
What the 43% figure actually represents
The 43% earnings growth projection for the technology sector covers a wide range of underlying businesses. It is not uniformly distributed.
At the high end, semiconductor companies supplying AI training and inference hardware — Nvidia being the paradigmatic case — are seeing demand that consistently exceeds supply. Data center customers (cloud hyperscalers: Microsoft, Google, Amazon, Meta) continue to invest aggressively in AI infrastructure. Their capital expenditure commitments for 2026 reached record levels, with the four largest hyperscalers collectively spending over $200 billion on data center infrastructure this year.
At the lower end of the growth range within tech are companies that are adjacent to AI but whose business models have not been transformed by it: legacy software vendors, IT services companies, hardware distributors. Many of these are growing earnings, but at rates far below the sector headline.
The 43% figure is a weighted average that obscures this dispersion. A passive technology ETF captures it all; a selective active approach attempts to concentrate exposure in the higher-growth components.
The AI market size projections
The AI market (broadly defined as AI software, AI-enabled hardware, and AI services) is projected to reach $3 trillion in total addressable market by 2033, according to various market research estimates (Bloomberg Intelligence, IDC, McKinsey Global Institute). This projection carries the standard caveats: it is a forecast of a technology whose commercialization trajectory is genuinely uncertain, and market size estimates in technology have historically been both dramatically underestimated and dramatically overestimated depending on the technology.
What is less speculative is the current year capital expenditure: the $200+ billion being spent by hyperscalers on data centers in 2026 is not a projection — it is disclosed in quarterly earnings reports. That spending is flowing through to revenues at Nvidia, AMD, TSMC, and dozens of semiconductor equipment companies.
The S&P 500's aggregate EPS growth projection for 2026 is 12.6%. Technology at 43% is growing more than three times faster than the broad market in earnings terms. The question is whether that earnings growth is already priced in — whether the stock prices of AI beneficiaries already reflect 43% earnings growth plus the next several years of continued growth.
What "second wave" actually means
The "first wave" of AI investing (roughly 2023–2025) was characterized by relatively indiscriminate buying of anything with AI exposure. Companies that added "AI" to their product descriptions saw stock price responses; pure-play infrastructure providers (Nvidia above all) became trillion-dollar companies in years rather than decades.
The "second wave" — which characterizes 2026 — involves a different set of questions:
Who actually monetizes AI? The chips and cloud infrastructure are being built. But for most enterprise software companies, AI features have been added to products without meaningfully increasing revenue per customer. The second wave asks: which companies are translating AI deployment into measurable revenue and margin expansion?
Where does the monetization actually show up? Legal tech, coding tools, healthcare AI diagnostics, financial data services — these are the sectors where AI is being embedded in workflows with demonstrable productivity effects. The revenue capture story is clearer here than in general-purpose AI platforms.
Energy as an AI co-beneficiary: Data centers require enormous amounts of electricity. The AI infrastructure buildout has created a power demand surge that is not widely appreciated in popular coverage. In 2026, the energy sector has emerged as an unexpected AI beneficiary — utilities and power generators that supply data center campuses are booking long-term contracts. This is one of the "pick-your-spots" themes that professional investors have identified as less crowded than direct AI hardware.
Brazil's AI exposure
Brazilian companies are not primary participants in the AI infrastructure buildout — the semiconductor manufacturing and cloud hyperscaler ecosystem is concentrated in the US, Taiwan, and South Korea. However, Brazilian companies are becoming AI adopters, and several are developing AI-native products for the domestic market.
In the fintech space, AI-driven credit underwriting, fraud detection, and customer service automation are widely deployed. Nubank has invested significantly in machine learning infrastructure. Brazilian insurance companies are using AI for claims assessment. This adoption creates productivity gains but does not yet create publicly traded Brazilian AI pure-plays at scale.
For Brazilian investors seeking AI exposure, the primary route remains IVVB11 (the B3-listed S&P 500 ETF) or BDRs (Brazilian Depositary Receipts) of US technology companies.
Risks that the 43% headline does not show
Concentration: AI earnings growth is heavily concentrated in a small number of companies. If Nvidia disappoints — if demand softens or a credible competitor emerges — the tech sector earnings figure falls dramatically. The index hides this concentration.
Regulatory risk: The EU's AI Act is in enforcement phase. US regulatory posture toward AI is evolving. Content licensing disputes (training data copyright cases) are working through courts. Any of these could create unexpected compliance costs or operational constraints for AI developers.
Geopolitical risk around semiconductors: US export controls on advanced chips to China have been a significant factor in the AI semiconductor market. Policy changes in either direction would affect the demand and supply dynamics of the AI hardware market materially.
The "picks and shovels" saturation: When every investment bank publishes a report recommending AI infrastructure picks, the obvious trades tend to be crowded. The market's more measured response in 2026 versus 2025 reflects in part that the most obvious AI beneficiaries were already well-owned.
An honest read of the data
The 43% earnings growth is real, and it is being generated by real capital spending on real infrastructure. The AI market is not a speculation without underlying economic activity — data center construction, chip sales, cloud revenue, and software contract renewals are all measurable in real time.
What is uncertain is the duration and distribution of returns. How many years does above-market tech earnings growth persist? Which second-order beneficiaries (energy, real estate for data centers, cooling equipment, networking) see durable gains? Which AI-adjacent companies see revenue impact that does not yet show up in current earnings?
These are the questions that distinguish the second wave of AI investing from the first.
At Royal Binary, our trading team monitors earnings delivery, sector rotations, and technical setups across US technology and Brazilian markets. If you'd like to understand how we approach data-driven market analysis, explore our platform.