Let me start with the number that should give everyone pause. The S&P 500 just posted its best month since November 2020, the month Pfizer announced its vaccine and the world exhaled. That month had a reason. A clear, binary catalyst that repriced the entire forward outlook in a single day. April 2026 is different. There is no resolution. The Strait of Hormuz is still blocked. Brent is above $125. Headline inflation has risen to 3.5%, its highest in nearly three years. US GDP grew at 2% in Q1, below expectations, with consumer spending slowing to 1.6% annualised. And yet the market posted a historic month.
The question worth asking is not whether the rally happened, which it did, but whether what is driving it is a signal or a distortion.
Tech is not a safe haven. It is being treated like one.
The framing I keep seeing is that investors are rotating into tech because it is defensive. And I understand the surface logic: these companies have pricing power, they are not rate-sensitive in the traditional sense, and their earnings are driven by structural AI demand rather than consumer cyclicals. When energy prices spike and consumer spending softens, the argument goes, where else do you put money?
State Street's custodial data shows institutional investors have been adding to US tech holdings since the conflict began. European and Asian funds have seen net outflows. $125 billion has poured into US equity ETFs in April alone. Citi upgraded US equities to overweight this month. Nomura's McElligott called it the "official killing off" of the brief dispersion out of megacap tech that started late last year.
All of that is real. But I think the "tech as defensive" framing conflates two things that should be kept separate: the quality of these businesses, which is genuinely high, and the price being paid for them, which is increasingly stretched. Alphabet up 34% in a month is not a defensive move. Intel up 114% in a month is not a defensive move. These are momentum-driven re-ratings, and momentum-driven re-ratings do not require bad news to reverse, they just require the pace of positive surprise to slow.
The earnings confirmed the thesis. They did not expand it.
The big four hyperscalers collectively reported 60% earnings growth and announced $725 billion in combined AI capex for the year, up 77% from last year's record. That is an extraordinary set of numbers, and it deserves credit. The AI investment cycle is real. The revenue conversion is beginning to show up, most visibly in Google Cloud (+63%) and Microsoft Azure (+40%). These are not speculative commitments to a future that may never arrive. They are funded by businesses that are already generating returns.
But here is what I think the market is glossing over. The earnings validated the existing thesis. They did not meaningfully raise the ceiling on it. Alphabet's results were excellent and the stock jumped 10% on the day, but the underlying story was already known: Google Cloud was growing, AI search was monetising, the capex was being deployed productively. What moved the stock was confirmation, not discovery.
Meanwhile, the cracks that did show up: Meta's user decline, rising memory costs flagged by Microsoft ($25 billion of its record capex budget attributed to component cost inflation), Hood's warning that cloud growth would remain constrained at least through 2026, are being discounted almost entirely. The market is selectively reading the earnings. Good news moves stocks. Bad news gets filed under "temporary."
"Every other region has seen capital leaving. Right now is the time to find areas which can withstand the squeeze on margins better than others, and for many, the answer is tech. What is the alternative?" - State Street
That last question, what is the alternative, is doing a lot of work in this rally. And it is a valid point. In a world where energy prices are squeezing European and Asian markets, where rate cuts have been pushed back, and where the safe-haven trade is increasingly crowded, US tech is the default destination for global capital. But "best available option" and "fairly valued" are not the same thing.
The macro divergence is widening, and that has a shelf life
The part of this story that I find most structurally interesting is the growing gap between what the tech index is pricing and what the broader economy is experiencing.
Consumer spending slowed to 1.6% annualised in Q1, down from 1.9% in Q4 2025. The personal savings rate is falling, some consumers are already spending more than they earn. Gas at $4 a gallon has not fully worked its way through spending data yet; Q2 will likely be worse. GDP grew at 2%, below trend, and the population growth tailwind that historically supports trend growth is fading.
None of this is catastrophic. But it is a set of data points that, in a normal market environment, would be prompting more caution. The reason they are not is that tech's earnings are structurally insulated from these dynamics in ways that consumer discretionary, industrials, and small caps are not. Interest rate sensitivity matters less when you have $460bn in cloud contract backlogs. Consumer spending weakness matters less when your customers are hyperscalers with multi-year capex commitments.
The divergence is real, but it creates a specific kind of fragility. The broader index is being carried by five to ten names. Goldman Sachs estimates 84% of high-growth tech stocks' enterprise value now derives from cash flows more than ten years out. A 1 percentage point change in long-term growth assumptions implies a 29% move in enterprise value. The market is pricing in a very specific and sustained future. The margin for error is narrower than the price action implies.
What I am watching
Morgan Stanley's Andrew Slimmon put it plainly this week: the rally looks stretched, positioning is aggressive, and a short-term pullback is more likely than not. I agree, and I think the catalyst, if it comes, is more likely to be a guidance disappointment than a macro shock. The macro risks are known and have been partially priced. What the market has not priced is the possibility that the AI revenue conversion story hits a speed bump, a quarter where capex guidance rises faster than cloud revenue, or where hyperscaler customers start scrutinising ROI on AI infrastructure more carefully.
Three things I am tracking into the second half:
Hyperscaler capex discipline. The $725 billion figure is impressive. But Microsoft already flagged component cost inflation as a constraint. If memory prices continue to rise and GPU availability stays tight, the cost side of the AI buildout could compress margins faster than the revenue side expands. Watch for any language around capex moderation in subsequent quarters, it would reprice the semiconductor names that have led this rally most aggressively.
Consumer spending in Q2. Gas prices above $4 were not fully reflected in Q1 data. A Q2 reading below 1% annualised consumer spending would be a different conversation. The Fed cannot cut into that environment if inflation is simultaneously running at 3.5%, and that stagflationary bind is what keeps me from being fully comfortable with the rally's durability.
The Warsh transition. The new Fed chair inherits a split committee, a president demanding lower rates, services inflation above target, and an oil shock of uncertain duration. His first meeting in June will be a signal. Either he reinforces the hold posture and manages the hawks carefully, or he introduces ambiguity that unsettles the rate path. Markets have not priced much uncertainty into the transition. They probably should.
The rally is not irrational. The businesses at its centre are genuinely excellent, the AI investment cycle is structurally real, and the relative case for US tech in a world of energy shocks and geopolitical fragmentation is sound. But a 15% monthly gain in the Nasdaq, with oil above $125 and consumer spending slowing, is a market telling you it has decided to look past a lot of things simultaneously. That works, until it stops working. And the things it is looking past have not gone away.
