Tuesday evening at 5:00 PM ET, Lisa Su walks into another earnings call — and this one is different.
For three years, the AMD vs. Nvidia conversation was lopsided in a way that made AMD bulls sound delusional. Nvidia owned the AI training market. AMD’s MI300 was a credible second source for inference, but nobody was building $1B clusters around it.
That changed in Q4.
The two deals that re-rate AMD
OpenAI — 6 gigawatts. A multi-generation build commitment around the MI450 GPU platform. The first 1 GW of deployment capacity goes live in 2026. To put 6 GW in context: that’s roughly the entire annual electricity consumption of New Zealand, dedicated to running AMD silicon. OpenAI is no longer hedging its compute supply chain. It’s actively building a second one.
Meta — $60 billion, five years. A 1 GW opening deployment on MI450 hardware, with options to expand. Meta has historically been one of Nvidia’s biggest customers; this is a structural shift, not a token order.
Together, these two deals provide AMD with revenue visibility through 2027 that didn’t exist six months ago. The data center segment posted $5.38 billion in Q4, up 39% year-over-year — a record. With these two contracts ramping, the segment trajectory through 2027 is no longer hostage to whether AMD can “win” a customer; it’s a question of execution on already-signed business.
What Wall Street wants tomorrow night
Consensus expects revenue of $9.84 billion, up 32% year-over-year, with EPS up roughly 33%. Gross margin guidance is 55% (non-GAAP). Those are the in-line numbers.
The number that actually matters is FY26 guidance. Consensus is sitting at $48 billion in revenue. If AMD guides to $50 billion or higher, the OpenAI/Meta deals are pulling forward into the model and the AMD/Nvidia revenue spread starts compressing — which is when AMD gets re-rated.
The Play
This is a high-bar setup. AMD is up 6% in the two weeks heading into the print, and the stock recently broke out above $300. Naked long shares into earnings against a stock that’s already extended is the wrong risk profile.
The cleaner trade: defined-risk call spreads. A May 9 $310/$330 call spread caps your loss at the debit paid (currently around $4) and pays out $20 if AMD rips. Implied vol on the May 9 chain is sitting near 65 — not cheap, but the structure protects you on a “good but not great” print where the stock fades from the open.
Watch the FY26 guide on the call. Above $50B = re-rating. Below $48B = the next two months are a chop fest.