The market is telling us something. Don’t ignore it.
Paid subscriber analysis
The market did something very important this week.
It looked at another episode of elevated war risk, oil getting back close to $100 (currently at $95, up from below $80 last Friday), higher headline inflation coming, political noise around the Fed, stretched positioning in semis, and a very heavy earnings calendar - and it still went higher.
Not because the world is suddenly safe. It’s not.
Not because there are no risks. There are plenty.
But because markets do not trade headlines. They trade underlying conditions. And the conditions are still pointing in one direction: earnings are accelerating, AI capex is not slowing, semis are leading, positioning is still not euphoric, and the market is beginning to look through the oil shock as long as the economy does not crack (we still have bonds warning us over that).
That was the message this week.
SPX pushed to 714 on SPY, while QQQ finished at 664, with Nasdaq strength led again by semiconductors and mega-cap tech. Nasdaq reached an intraday record, driven by a powerful semiconductor rally after Intel’s results. The PHLX Semiconductor Index extending its rally to an extraordinary 18 consecutive sessions.
That is not a defensive tape.
That is a tape telling you investors are still willing to pay for growth, especially when that growth is tied to AI infrastructure.
It’s like the whole AI killing SaaS and consequentially private credit/private equity was shrugged as easily as oil prices and the Iran war.
Just a reminder, I wrote about that recently, if you missed it here it is again:
On the other hand, VIX closed at 18.7 with oil prices at $95. Three weeks ago I pointed this decoupling as a likely market bottom. Oil is still elevated, inflation risks are still there, but the market is not breaking a sweat over it.
As mentioned last time, earnings can be the only thing that disturb/break this momentum. And they delivered.
Earnings: AI demand is still the only story that matters
According to FactSet, with 28% of S&P 500 companies having reported Q1 results, 84% have beaten EPS estimates and 81% have beaten revenue estimates. That is well above the 5-year and 10-year averages on both lines. More importantly, the aggregate EPS surprise is +12.3%, versus a 5-year average of +7.3% and a 10-year average of +7.1%. This is not a weak earnings season being saved by a few accounting tricks. This is a real upside surprise.
Check it out, across every sector:
But the most important earnings report this week was not Tesla. It was Intel.
Intel reported first-quarter revenue of $13.6 billion, up 7% y-o-y, and Q2 revenues to $14.8 billion, ahead of expectations. The market reaction was violent because the report hit exactly where the market was most nervous: AI infrastructure demand. Intel’s data center and AI segment reportedly delivered $5.1 billion in revenue, up 22% year over year, well ahead of expectations.
This matters because it directly challenges the “AI overbuild” narrative.
For months, bears have been saying that the hyperscalers are spending too much, too fast, on too much capacity, for too little return. That may still become true one day. But the evidence this week went the other way. Intel said demand was strong. Supply was constrained. The market heard the same thing it has heard from Nvidia, Broadcom, AMD, and the hyperscalers: the bottleneck is not demand. The bottleneck is capacity.
That is why Intel did not just help Intel. It helped the whole AI complex on Friday. AMD rallied sharply after Intel’s results, with analysts interpreting the numbers as evidence of broader strength in CPU demand tied to AI inference workloads. Nvidia was also up strongly on Friday, trading around $209, up nearly 5% intraday.
This is the key point: the market is no longer just rewarding AI dream stories. It is rewarding companies that can show AI demand in the P&L.
Tesla was more complicated.
Tesla reported this week, and the market’s reaction was muted because investors are still trying to separate the auto business, the AI/robotics story, the margin recovery story, and the regulatory credit issue. One of Tesla’s cleanest profit streams - regulatory credits - is shrinking. Tesla earned $380 million from regulatory credits in Q1, down from $542 million in Q4, and future performance obligations have fallen sharply.
So Tesla is still not an earnings story in the same way Microsoft or Meta. Tesla remains a duration asset wrapped in an auto company wrapped in an AI optionality trade. That can work. But it requires the market to believe in the future more than the present.
Overall, the earnings message this week was not “everything is fine.”
It was more specific than that.
AI demand is fine. Semis are fine. Cloud and infrastructure are likely fine. The consumer is not broken. But oil is becoming a margin problem for companies that cannot pass it through.
Next week: the real earnings season begins
Next week is the real test.
