Microsoft just delivered a masterclass in “close, but no cigar” by quietly pushing back the launch of its next-gen Maia AI chip (codename Braga) from 2025 to at least 2026. The reason? A cocktail of design chaos, developer churn, and the brutal reality that their silicon still can’t keep up with Nvidia’s Blackwell. In fact, if you listen closely, you can hear Jensen Huang lighting another celebratory cigar with a $100 bill.
(Source: X)
In short, Microsoft has been trying to build a homegrown AI chip since 2019, mostly to stop hemorrhaging cash to Nvidia every time someone in Redmond says “Copilot.” The whole point of Braga was to finally cut the cord… a.k.a. run their own AI workloads, save a few billion, and maybe even flex on Amazon and Google, who already have their own (actually functional) chips in the wild. Instead, they’re stuck in the world’s most expensive group project, and the only thing they’ve built so far is tracing paper and a pile of resignation letters.
(Source: Tom's Hardware)
Case in point: the Maia 100 chip, which Microsoft hyped last year as a “game-changer,” is apparently being used for internal testing and not much else. Why? Because it was designed before ChatGPT turned “large language models” into the only thing anyone cares about, and it’s optimized for image processing, not the generative AI arms race that’s currently setting fire to every data center budget on Earth. Translation: It’s like showing up to the Indy 500 in a Segway.
Meanwhile, Amazon’s Trainium3 and Google’s seventh-gen TPUs are already out here kneecapping Microsoft’s hardware left, right, and twice on Sunday. Meanwhile, as Satya’s team is busy putting out fires, Nvidia is just standing there, arms crossed, watching the competition eat glue. Huang even said out loud that most of these “custom chip” projects will end up in the dumpster. So far, Microsoft seems determined to prove him right.
(Source: Giphy)
Financially speaking, the pain is real AF. Microsoft’s cloud margins are getting squeezed every quarter they’re forced to buy Nvidia hardware at “bend over” prices. Sure, Copilot might add $25 billion in revenue by 2026 (if you believe Wedbush’s wet dreams), but every time the Maia project slips, that number looks more like Monopoly money. Their stock barely budged (read: -0.30% on Friday) on the news, presumably because the market is too busy chasing interest rate foreplay and “deadline’s aren’t critical” sermons to care about the fact that Microsoft’s silicon roadmap is legit being held together with a prayer.
Meaning, as it stands right now, the dream of breaking free from Nvidia is looking about as realistic as George Strait joining Beyoncé on her “Cowboy Carter” tour. Spoiler: Never going to happen. For now, all roads… and all margins, continue to lead straight through Jensen Huang’s front door. Of course, only time will tell if Microsoft actually gets their sh*t together, but until then, keep your eyes on this story and place your bets accordingly. Until next time, friends…
At the time of publishing, Stocks.News holds positions in Microsoft, Amazon, and Google as mentioned in the article.
Did you find this insightful?
Bad
Just Okay
Amazing
Disclaimer: Information provided is for informational purposes only, not investment advice. We do not recommend buying or selling stocks. Stock price discussions are based on publicly available data. Readers should conduct their own research or consult a financial advisor before investing. Owners of this site have current positions in stocks mentioned thru out the site, Please Read Full Disclaimer for details Here https://app.stocks.news/page/disclaimer