Continue Reading With Our 7-Day Free Trial
Hey there. If you were watching your screens Wednesday morning, you probably saw Meta Platforms (NASDAQ:META) rip more than 9% higher before markets even settled into the day. That’s a massive single-session move for a company with a market cap north of $1.5 trillion. The trigger? A Bloomberg report that Meta is building a cloud business to sell its excess AI computing capacity — a move that could put it in direct competition with Amazon, Microsoft, and Alphabet.
Here’s the twist, though. The obvious read is “Meta finally figured out how to make money on its AI spending.” But the more interesting story is what happened everywhere else. Chip stocks got hammered. Neocloud players got crushed. And underneath it all, there’s a bigger question about whether the entire AI trade is quietly shifting shape. Let’s break down what’s actually going on.
Meta’s Excess Compute Bet Rattles Neoclouds
Let’s start with the headline. Bloomberg reported Meta is weighing two paths: hosting AI models for outside developers (similar to Amazon’s Bedrock) or renting raw GPU capacity the way neocloud providers like CoreWeave have built their entire business around. Either way, Meta would be walking straight into a market currently dominated by AWS, Azure, and Google Cloud.
The market’s reaction was immediate and brutal for a specific corner of the trade. CoreWeave and Nebius Group both plunged around 12% on the news. That’s not a coincidence. Both companies hold multibillion-dollar supply contracts with Meta, meaning the company that was buying their capacity might soon be competing with them for the same customers.
There’s real irony here, folks. Meta has spent the better part of the last year scrambling for enough compute to train its models — at one point, Google reportedly had to ration Meta’s access to its own Gemini models because it couldn’t spare the capacity. Now Meta is positioning itself to sell the overflow. That tells you something about just how lumpy and front-loaded this entire infrastructure buildout has become.
Meta hasn’t officially confirmed the plan yet, and the report notes the strategy could still change. But CFO Susan Li flagged the idea back on the Q1 2026 earnings call, saying cloud services were “on the table” if Meta built more capacity than it needed. This isn’t coming out of nowhere.
The Chip Selloff Nobody Was Expecting Mid-Rally
While Meta was celebrating, the chip sector was having a very different Wednesday. The PHLX Semiconductor Index slipped roughly 4% in the opening hour, a sharp reversal after posting its best quarter on record with a first-half gain of more than 100%. Micron, AMD, Marvell, and Intel all took meaningful hits.
Why does a cloud announcement from a social media company move chip stocks? Because Meta signaling it has excess compute implies the most urgent, panic-buying phase of the AI infrastructure race might be cooling off. If the biggest spenders are shifting from “buy everything” to “let’s monetize what we already have,” that’s a read-through the market can’t ignore.
It’s worth remembering the scale we’re talking about. Big Tech is expected to spend more than $700 billion on AI infrastructure this year, up from roughly $400 billion in 2025. Meta alone raised its 2026 capex guidance to a range of $125 billion to $145 billion, driven largely by higher component costs, especially memory pricing. When a spender that large hints at having more than it needs, chip investors take notice.
Meanwhile, there’s a genuine rotation happening. Software stocks, which had been left for dead earlier this year, finally found buyers. The iShares Expanded Tech-Software ETF rose 3.2%, helped by upgrades on Salesforce and ServiceNow from Guggenheim, whose analyst argued the “Armageddon scenario” priced into software stocks didn’t match reality. Money is moving, not necessarily leaving.
Why Meta’s Spending Finally Has A Revenue Story
For months, Wall Street has been uneasy about Meta’s capital spending without a clear payoff attached to it. This cloud plan changes that framing, at least on paper. Instead of AI infrastructure being purely a cost center, it becomes something that could generate direct external revenue — a meaningfully different narrative for a stock that had been under real pressure.
And Meta has been under pressure. Despite Wednesday’s pop, the stock remains down more than 22% from the highs it hit last summer, and it was still down close to 15% year-to-date heading into this news. This wasn’t a stock riding high looking for an excuse to go higher. It was a laggard among the Magnificent Seven looking for a reason to catch up.
The fundamentals underneath are still strong, worth remembering. In Q1 2026, Meta posted total revenue of $56.3 billion, up 33% year-over-year, with ad revenue growing at the same clip. Operating income came in at $22.9 billion, a 41% operating margin. Free cash flow was $12.4 billion. This is not a company struggling to generate cash — it’s a company debating what to do with the compute it’s already built.
CEO Mark Zuckerberg has also been clear that efficiency matters alongside growth. Meta reduced headcount earlier this year and has been leaning on AI internally to boost engineering output. A cloud business, if it materializes, would be one more lever to pull toward turning that $145 billion capex number into something investors can actually underwrite.
What This Means For The Broader AI Trade
Here’s where it gets genuinely interesting, and it’s worth sitting with for a second. Freedom Capital Markets’ Jay Woods put it well: “The Magnificent Seven aren’t broken, they are evolving.” For the first time since the AI boom kicked off, investors appear to be rotating away from companies purely spending on AI and toward companies figuring out how to build the ecosystem around it.
That distinction matters more than it sounds. Saxo Bank’s Charu Chanana flagged the real risk facing the AI trade heading into the back half of the year: it’s not that AI demand disappears, but that demand becomes more price-sensitive just as markets have been pricing it as nearly unlimited. Microsoft reportedly canceled some Claude Code licenses last month while reviewing operating expenses — a small data point, but a telling one.
China’s AI advances are adding to that uncertainty too, with companies increasingly focused on token costs as agentic AI adoption expands. If token prices fall sharply, usage could improve, but that might squeeze revenue and margin expectations for some AI infrastructure providers at the same time.
The tech trade has been carrying the entire market. It’s been the sole contributor to S&P 500 earnings growth over the trailing twelve months and is expected to account for roughly 75% of second-quarter profit growth once this reporting season wraps. That’s an enormous amount of weight resting on a handful of names — which is exactly why a single Bloomberg report can move the whole board like this.
Final Thoughts
So, where does that leave Meta specifically? On valuation, the picture has actually gotten cheaper over the past year, not more expensive, even with Wednesday’s pop. Meta’s NTM Price/Earnings multiple sits at roughly 18.8x, down from around 29x a year ago. Its NTM EV/EBITDA multiple of about 10.3x and NTM EV/Revenue of 5.9x are also well below where they stood in mid-2025. On a trailing basis, LTM P/E sits near 22.3x and LTM EV/EBITDA near 14.3x — multiples that don’t scream excess relative to Meta’s own history, even after this week’s rally.
That said, a couple of numbers are worth flagging without over-interpreting them. Meta’s NTM levered free cash flow yield has actually turned negative, a byproduct of the aggressive capex ramp eating into near-term cash generation, and something to watch as spending continues into 2027. The market seems to be underwriting the idea that this spending eventually converts into revenue — whether through better ads, new agentic products, or now, potentially, a cloud business.
Whether that conversion happens on the timeline investors are hoping for is genuinely an open question, not a foregone conclusion either way. For now, the multiples suggest Meta isn’t priced for perfection the way some of its AI-infrastructure peers are — but the negative free cash flow yield is a reminder that the payoff on all this spending still has to show up. Worth keeping on your radar as we head deeper into earnings season.
Disclaimer: We do not hold any positions in the above stock(s). Read our full disclaimer here.
