Nvidia, the Hollywood industrial complex
MEMO: To Everyone Still Pretending This Makes Sense
Internal correspondence from someone who has been watching the tape too long
Friday, February 27, 2026
RE: Nvidia, the Hollywood industrial complex, PPI, and the general collapse of narrative coherence
FROM: The desk of a person who opened the Bloomberg terminal at 6 a.m. and has not eaten since
TO: Anyone still capable of forming a coherent investment thesis
CLASSIFICATION: Urgent / Slightly unhinged
Let me begin with the most clarifying data point of the week.
Netflix — the company that spent months positioning itself as the savior of Warner Bros. Discovery's soul, the strategic acquirer, the streaming colossus that would unite HBO's prestige library with its own scale — walked away from its $83 billion deal on Thursday night. Just. Walked. Away.
The co-CEOs released a statement so financially disciplined it almost read as performance art. Warner Bros., they said, was "always a 'nice to have' at the right price, not a 'must have' at any price." Just like that. Two months of negotiations, Paramount lobbing hostile tender offers directly at shareholders, sovereign wealth funds from Saudi Arabia, Abu Dhabi, and Qatar quietly financing the other side, Netflix co-CEO leaving a White House meeting to announce the end of it all — and the takeaway is: nice to have. Not at any price. Fine.
So now Paramount Skydance walks away with Warner Bros. Discovery at $31 per share, the WBD board has declared it "superior," and Netflix will receive a $2.8 billion breakup fee as a parting gift — which caused Netflix shares to surge more than 10% in after-hours. Read that back. The company that lost the deal made more money on the news than the company that won it.
This is the market. This is 2026. The scoreboard is decorative.
Meanwhile, Paramount absorbs what could be — depending on which banker you believe — $6 billion in "synergies." Netflix's own chief global affairs officer said publicly that this number, which is one of the more bloodless words in the corporate vocabulary, would arrive primarily in the form of layoffs. An unspecified and truly spectacular number of people employed by two of the oldest studios in Hollywood are now waiting to find out whether their job survives the merger of Paramount Skydance and Warner Bros. Discovery, a combined entity that controls HBO, CNN, Paramount Pictures, CBS, Nickelodeon, and enough IP to make a content landfill. The integration will be baroque. The redundancies will be methodical. The press releases will describe it as "exciting."
I have no advice here. Only a kind of exhausted awe that the entertainment industry, having spent five years telling us the streaming wars were over, has now consolidated into a structure so large it will require three antitrust attorneys and a constitutional scholar to assess.
Now. Nvidia.
The company printed fiscal fourth-quarter EPS of $1.62 against a consensus of $1.53. Revenue: $68.13 billion, versus $66.1 billion expected. Forward guidance: $76.44 billion to $79.56 billion, against an analyst expectation of $72.6 billion. Every number was wrong in the right direction. By every traditional metric, this was a good quarter.
The stock fell. Thursday it dropped again. Down another 3.24% with Broadcom, Micron, Intel, and Applied Materials falling between 5% and 7% in sympathy, turning what should have been a sector-wide exhale into another session of passive chipmaker obliteration.
Here is what I think is actually happening, and I want to be precise about it: the AI capex story is losing its grip. Not because the spending has stopped — it hasn't, not yet — but because the market's ability to extrapolate that spending infinitely into forward earnings is visibly fraying. The hyperscalers are building. They're also increasingly designing their own chips. Sovereign AI programs across Europe and Asia are standing up their own compute infrastructure. And the inference layer — the place where Nvidia's bulls see the next supercycle — remains mostly theoretical at the revenue line for anyone outside of a handful of frontier labs.
The quarter was fine. The question underneath the quarter is not fine. The question is: what happens to demand if capital expenditure discipline becomes fashionable again? What happens to the narrative if one major hyperscaler decides to spend less in Q3 and makes a point of saying so on their earnings call?
Nobody wants to hold through that answer. So the stock reacts the way it did — not catastrophically, just... persistently downward, in the direction of a thesis being tested against reality.
Salesforce went up 3% on a "lukewarm outlook." Relative performance is a strange religion.
The PPI data drops today, February 27, at 8:30 a.m.
The context matters. January CPI came in at 2.4% year-over-year, running hotter than comfortable. The Fed government shutdown — yes, this happened, it delayed the BLS's release schedule by nearly two months — means we're receiving data whose timeliness has already been compromised. What we're getting this morning is January's producer price picture, viewed in late February, while March 4 tariffs on Canada and Mexico sit on the calendar like an undetonated charge.
Here is the sequence that should concern you: CPI is already running warm. PPI leads CPI. Tariffs are going live in less than a week. If today's print surprises to the upside — if it echoes the December data, where headline annual PPI held at 3% and core PPI accelerated to 3.3%, the highest in months — then the Fed's four-cuts-in-2026 trajectory starts to look like something drafted before anyone ran the numbers on import costs from North America.
The construction of this scenario is not complicated. It is in fact so simple it should embarrass everyone who built a rate-cut trade on the assumption that tariffs were mostly a negotiating tactic and inflation had been tamed. Neither thing is confirmed. One of them — the tariff thing — is about to become either very confirmed or very unconfirmed by Tuesday morning. You will know by the time you read your inbox on March 5 whether the administration decided the leverage was worth the inflation or decided the inflation was worth the climb down.
Place your bets accordingly. Or don't. I've watched too many confident trades die in the gap between policy intent and implementation.
One more thing before I go in search of food.
The S&P 500 closed at 6,946 on Wednesday after a two-day rally that felt, in the words of someone far calmer than me, like a market approaching "the top of its near-term 6,800 to 7,000 trading range." The VIX fell 8.3% to 17.93. Volume was light — 17.5 billion shares against a 20-session average of over 20 billion. Thin rallies in compressed ranges with six of eleven S&P sectors still in the red. This is not the internal structure of a market that has made up its mind. XLK and XLF carried the index. Industrials sold off. CAT dropped 3.3%.
The industrials tell you something the tech rally is trying not to say: that somewhere in the real economy, the tariff math is already running. Caterpillar sells heavy equipment. Heavy equipment gets purchased during infrastructure buildouts. Infrastructure buildouts get deferred when cost inputs are uncertain. The stock fell 3.31% Thursday. That's not a rounding error — it's a signal wearing a work boot.
Anyway.
Netflix gets $2.8 billion to walk away from something it described as a nice-to-have. Nvidia beats estimates and falls. Paramount absorbs Warner Bros. Discovery and begins the process of figuring out which half of Hollywood it needs. PPI lands this morning with everyone quietly hoping it doesn't invalidate whatever macro framework they built their portfolio around. And March 4 is still coming.
The week was not boring. I will give it that.