The successful $5.5 billion IPO of a semiconductor startup last week provided a striking indication of just how hot the sector has been.
The up-sized transaction, valuing the company at $40 billion, was reportedly 40 times subscribed, highlighting the depth of investor demand for AI-related and semiconductor shares that has driven one of the most extraordinary rallies in recent memory.
The S&P 500's semiconductor companies alone have added about $3.8 trillion in market capitalization over the past six weeks, part of a year-to-date surge that has seen the SOXX, the benchmark ETF for U.S.-listed semis, gain 71.55% through May 13.
That demand has also helped deliver an equally extraordinary period for U.S. equities more broadly. April opened with real headwinds: spiking Treasury yields, evaporating Federal Reserve cut expectations, and a tariff environment pushing investors into defensives while growth names sold off sharply. Yet what turned the tide was a blowout first-quarter earnings season that reignited conviction in the AI investment cycle.
The result was the S&P 500's best monthly gain since November 2020, recovering more than 16% from its March lows.
Earnings Plus Evidenced AI Demand
The fundamental backdrop, in our view, justifies the attention and potentially bodes well for other AI IPOs to come. With 89% of S&P 500 companies having reported actual results (as of May 8), first-quarter blended earnings growth came in at 27.7%—the strongest since Q4 2021—with 84% reporting a positive earnings surprise. In aggregate, companies beat estimates by an average of 18.2%, the highest percentage since Q1 2021, and nearly three times both the five- and 10-year norms. Hyperscalers Alphabet, Amazon and Meta were the three largest positive surprise contributors, fueling a sharp rotation back into growth. Technology reclaimed market leadership decisively, with 10 of 11 S&P 500 sectors reporting year-over-year earnings growth.
What has fundamentally shifted the semiconductor narrative, however, is not just earnings beats—it is hard evidence that real-world AI demand is materializing faster than anticipated. That distinction matters. Leading semiconductor companies had strong visibility into 2027 well before this rally took hold, yet investors remained skeptical, focused on the unresolved question of AI's return on investment.
What changed that calculus was hard revenue proof points of a kind that are unprecedented. Our analysts have estimated that from annual recurring revenue of $9 billion at the end of 2025, AI startup Anthropic's revenue grew to $30 billion by April 2026—a more than threefold increase in just four months, equivalent to $1 – $2 billion of new revenue every single week. Visibility into 2027 spending is now largely locked in and has been revised upward; 2028 is where genuine debate remains.
A Broader Opportunity Set
For years, the semiconductor narrative revolved almost entirely around training large language models. That bifurcation is now unwinding, driven by the shift to agentic AI—applications that infer what users need and automatically perform tasks at unpredictable scale—which requires vast fleets of general-purpose components working alongside cutting-edge accelerators.
Three groups are emerging as clear beneficiaries. Memory remains a major area of strength, underpinned by structural supply constraints and firm pricing. CPUs are experiencing a meaningful re-rating; agentic AI requires a centralized orchestration layer to direct actions across accelerators and applications, a function that falls to the CPU. Analog semiconductors are being drawn into the story through the power devices that data centers require in volume.
Beyond chips, the agentic shift is reinvigorating demand for components deeper in the IT stack, including optical transceivers and multilayer ceramic capacitors, a development we explored in our recent piece, Hardly Boring: Agentic AI Is Reinvigorating the Legacy Hardware Sector.
Cautious But Confident
We have grown more constructive on the semiconductor sector as the evidence has accumulated, but the rally's composition deserves scrutiny. The sharp move in the SOXX has not been a rising tide; it has actively pulled capital from elsewhere.
Bank stocks are a case in point. Despite positive earnings revisions for 2026 and 2027, financials have been essentially flat since the Q1 earnings season began, even as the SOXX surged roughly 30%. The dynamic is partly mechanical—mutual funds underweight semis forced to rebalance—and partly rate sensitivity and risk-on/risk-off dynamics. Compounding the pressure on financials is the pending Senate Banking Committee markup of the CLARITY Act, legislation that would legitimize stablecoins and open payment and custody services to non-bank entities.
Near-term platform consolidation is also worth watching. Enterprises are trialing multiple AI platforms simultaneously; that duplication will rationalize. Our base case is a multi-platform outcome where several ecosystems coexist, sustaining rather than undermining semiconductor demand, but the smoothness of that transition matters.
Momentum indicators are at extreme historical levels. Just 28 names account for 50% of the S&P 500’s returns (to May 6) from the end of March, and individual stocks can pull back 20% in days. Those episodes, absent genuine deterioration in the AI infrastructure build, are opportunities rather than exits.
So, is the semis surge sustainable? The fundamentals say yes. AI demand is real, adoption is accelerating and spending visibility extends well into 2027. The path will not be linear, but the direction of travel is clear.
What to Watch For
Monday 05/18:
- Japan GDP
Wednesday 05/20:
- Eurozone Consumer Price Index
- U.K. Consumer Price Index
Thursday 05/21:
- U.S. Philadelphia Fed Manufacturing Index
- U.S. Manufacturing Purchasing Managers’ Index
- U.S. Services Purchasing Managers’ Index
Friday 05/22:
- Germany GDP