A Ceasefire, a Chip Deal, and the Fastest Mood Swing in History

On April 7, tech stocks couldn't decide if the world was ending. On April 8, they decided it wasn't. The difference was one phone call between Washington and Tehran — and about sixteen dollars off a barrel of oil.

I don't write about markets much. I'm a developer, not a trader, and most financial commentary reads like astrology with better formatting. But what happened on April 7 and 8 was worth paying attention to — not because the numbers were big (they were), but because of how quickly a war, a chip deal, and an earnings report all collapsed into the same 48-hour window. And what that tells you about where the actual power sits in tech right now.

Monday: Everything Happens at Once

Broadcom jumped 6.21% on April 7. That's a huge single-day move for a company worth over $800 billion, and it happened because of a deal that most people outside of infrastructure never heard about: Broadcom confirmed an expanded partnership with Google and Anthropic to supply custom TPU chips — 3.5 gigawatts of compute capacity starting in 2027. Mizuho estimates Broadcom's AI revenue from this deal alone at $21 billion in 2026 and $42 billion in 2027.

I'll say that differently. One chip contract with two customers is projected to generate more revenue than most Fortune 500 companies make total.

The same day, Samsung reported preliminary earnings with an operating profit roughly eight times what it made the year before. KI memory chips. That's it. That's the story. Samsung isn't growing because phones are selling better. It's growing because the AI infrastructure buildout requires an amount of high-bandwidth memory that didn't exist three years ago.

Apple went the other direction. Down 2.07% on reports of engineering problems with the foldable iPhone — specifically the crease, the hinge durability, and the display coating. By Tuesday, Bloomberg confirmed the September launch is still on track. The stock recovered. But that brief dip was a reminder: Apple is the one mega-cap that doesn't have a compelling AI narrative yet. When the market is rewarding AI spending and punishing everything else, not having that story costs you.

ASML dropped on new U.S. export restrictions targeting China. This is becoming a pattern — every few weeks, the Commerce Department tightens the screws, ASML's stock takes a hit, and then everyone moves on until the next round. The long-term issue is real, though: if China can't buy EUV lithography machines, the semiconductor supply chain fractures further. And the Iran war already fractured it plenty.

Tuesday: The Ceasefire Trade

Then Trump and Iran agreed to a two-week ceasefire.

The Strait of Hormuz reopened. Oil collapsed roughly 16%, from around $113 to the mid-$90s. And the markets lost their minds. Dow futures jumped over 1,000 points. Nasdaq-100 futures surged 3.4%. South Korea's Kospi popped 6.87%. The DAX rose about 4.7%. Samsung gained another 7.12%. SK Hynix — which makes the HBM chips that Nvidia can't get enough of — jumped 9.61%.

Goldman Sachs reportedly called the tech sector the best buying opportunity of the year, noting that tech's P/E ratio had fallen below consumer discretionary for the first time in a while. That framing is doing some work. It implies that tech was cheap relative to, say, luxury goods companies — which says less about tech being undervalued and more about how distorted the whole market got during the war.

Bitcoin sat at $68,166 on April 7, basically flat. Gold was at $4,654. Neither moved much on the ceasefire news, which is notable — usually one of them does.

What the Ceasefire Doesn't Fix

Here's the part the rally priced in too quickly.

Iran's missile strikes on Qatar's Ras Laffan complex in late February destroyed roughly 30% of the world's semiconductor-grade helium supply. That's not a number you fix with a ceasefire. Repairs are estimated at three to five years. Helium is essential for chip fabrication — it's used in lithography cooling and wafer processing. There is no substitute at scale.

Bromine is the other problem nobody's talking about. Israel and Jordan produce about two-thirds of the world's supply. Bromine is a key input for photoresist chemistry — without it, you can't pattern circuits onto silicon. The war disrupted shipping routes, and even a ceasefire doesn't instantly restart supply chains that have been operating in crisis mode for six weeks.

Taiwan's semiconductor industry association — TSIA — publicly asked the government on the same day to build strategic reserves of helium and liquefied natural gas. When TSMC's industry group is asking the state to stockpile raw materials, the supply chain isn't just strained. It's structurally damaged.

Full recovery, even in a best-case peace scenario, is estimated at four to six months. The market priced in relief in four to six hours.

The Musk-Altman Escalation Nobody Asked For

While the markets were celebrating peace, Musk and Altman were escalating their war.

On April 7, Musk's legal team filed amended claims in the OpenAI lawsuit demanding Sam Altman's removal as board director and officer, plus a full reversal of the for-profit conversion. The twist: Musk says any damages should go to OpenAI's nonprofit arm, not to him personally. That's a clever framing — it makes the suit look altruistic rather than competitive, which is exactly what you'd do if you were also running xAI and trying not to look like you're suing a rival for commercial advantage.

OpenAI countered the same day by asking the attorneys general of California and Delaware to investigate Musk's "improper and anti-competitive behavior." They specifically alleged Musk is coordinating with Zuckerberg against OpenAI. Jury selection starts April 27.

I don't have a prediction on who wins. I have a prediction that whoever loses will appeal, and this will still be going in 2028. What I do know is that the combined legal spending between these two could probably fund a mid-sized AI lab.

The Real Story: Three Companies Control the AI Supply Chain

Step back from the daily moves and there's a structural picture emerging that I think matters more than any single stock.

Anthropic is now at a $30 billion annualized revenue run rate. More than 1,000 enterprise customers paying over $1 million each — doubled since February. They just locked in 3.5 gigawatts of TPU compute through the Broadcom-Google deal. Broadcom's stock jumped 6% on that news alone.

Intel announced it's joining Musk's $25 billion Terafab project — a chip megafactory for Tesla, SpaceX, and xAI. Intel will help convert chip designs for mass production. The stock rose about 4.2%. For Intel's struggling foundry business, this is a lifeline. For Musk, it's another step toward vertical integration of his entire AI stack.

And Hermeus raised $350 million at a $1 billion valuation to build unmanned hypersonic fighters. Eclipse Ventures closed a $1.3 billion fund for AI infrastructure and robotics. Defense tech keeps absorbing capital at a rate that makes consumer AI funding look modest.

The pattern: compute, chips, and infrastructure are consolidating around a handful of players — Anthropic, Google, Nvidia, Broadcom, Musk's constellation of companies. The companies that control the silicon and the power supply are pulling away from everyone else. Broadcom doesn't care whether your startup makes it. They care whether Anthropic and Google keep buying TPUs. And right now, the answer to that question is a multi-decade contract worth tens of billions.

If you're building a company that depends on this infrastructure, you're a tenant. And the landlords just got a lot richer.