TL;DR
Meta’s stock rose about 6% on Friday and roughly 15% on the week, its best week since early 2024, after the company gave investors a concrete plan to monetise its AI infrastructure via Meta Compute. Wolfe Research estimates each gigawatt monetised at a ~$25bn rate could lift EPS around 20%, and options volume ran over three times its 30-day average. But Meta Compute has not sold anything yet, has never served external cloud customers, and faces three entrenched hyperscalers.
Meta has had a miserable year on the market, flat while the Nasdaq-100 climbed 18%. That changed abruptly, with the stock posting its best week since early 2024, CNBC reports.
Shares rose about 6% on Friday and roughly 15% across the week. The move was not driven by advertising, the business that actually makes Meta’s money.
It was driven by a story about compute.
What changed
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
The catalyst is Meta Compute, a plan to sell AI computing capacity and models to outside customers. It would put Meta into a market currently held by Amazon, Microsoft, and Google.
Investors had been anxious for months about the scale of Meta’s AI capital spending with no visible route to a return. The company has now offered to rent out its spare AI compute, and the market treated it as relief.
Mark Zuckerberg had already signalled the direction, saying an AI cloud business “makes sense”. Turning a colossal cost centre into a revenue line is a straightforward pitch.
The supporting pieces landed at the same time. Meta shipped Muse Image, its new image model, and is pushing its own MTIA chip into production to cut its dependence on Nvidia.
How the bulls are doing the maths
The numbers being circulated are aggressive. Wolfe Research reckons that for every gigawatt of compute Meta monetises at roughly a $25bn rate, earnings per share could rise around 20%.
Options desks responded in kind. Volume ran at more than three times the 30-day average, with 78% of the $1.8bn in options premium tied to calls.
Analyst targets have followed the mood, clustering in the low-to-mid $800s over twelve months. The published scenario range runs from about $720 at the bearish end to roughly $869 at the bullish one.
Those are forecasts, not facts, and they are being made by people who were considerably less enthusiastic a month ago. Sentiment moved before any of this revenue exists.
The part the rally is ignoring
Meta Compute has not sold anything yet. Meta has never run a cloud business for external customers, and AWS, Azure, and Google Cloud have a decade of head start on operations, sales, and trust.
There is also an uncomfortable reading of the same news. Selling excess capacity can mean shrewd monetisation, or it can mean the company bought more compute than it can use.
And the human ledger has not improved. Meta cut 8,000 jobs while posting record revenue and pouring money into AI infrastructure, which is the same spending this rally is celebrating.
What Wall Street bought this week was a narrative, and a plausible one. Whether Meta can actually sell compute against three entrenched hyperscalers is a question no share price can answer yet.
