Intel’s stock has climbed 490% in the past year. The company’s actual turnaround? Barely started. That gap is the real story, according to TechCrunch AI’s read of Bloomberg’s deep dive into CEO Lip-Bu Tan’s rescue mission at the most storied chipmaker in Silicon Valley.
Tan took the job in March of last year. TechCrunch AI reports he’s spent most of his first year doing deals rather than restructuring the business. The deal list is impressive on paper: a sweetheart arrangement that made the U.S. government Intel’s third-largest shareholder, a factory partnership with Elon Musk, and preliminary manufacturing agreements reportedly in motion with both Apple and Tesla.
That’s the bull case Wall Street is buying. Now look under the hood.
The fundamentals tell a different story
Intel’s chip yields still lag well behind TSMC, the industry leader everyone is trying to catch. Employees told Bloomberg that Tan has been light on internal specifics. Some teams are quietly adjusting missed deadlines rather than recovering from them. That’s a tell. When engineering orgs start moving the goalposts instead of hitting them, the rot is operational, not strategic.
What stands out here is the mismatch between narrative and execution. A 490% rally typically signals that investors believe the worst is over and growth is locked in. Intel’s situation is closer to the opposite: a credible CEO has bought the company time and political cover, but the actual manufacturing problems that knocked Intel off its perch are still unresolved.
Why this matters now
Three dynamics are colliding:
- Geopolitics is rewriting chip economics. Washington wants a domestic foundry champion badly enough to take an equity stake. That’s a structural tailwind no other U.S. chipmaker has.
- Hyperscaler demand is desperate. Apple, Tesla, and anyone training frontier models needs alternatives to TSMC’s capacity ceiling. Intel doesn’t need to beat TSMC to win business. It just needs to be good enough and available.
- Execution risk is the whole game. Yields are physics, not press releases. You can’t deal-make your way to a working 18A node.
This is the pattern to watch across the AI hardware stack right now. Capital is flowing toward anyone who can credibly promise non-Nvidia, non-TSMC capacity. The bet isn’t that these challengers will be best. It’s that they’ll exist.
What practitioners should take from this
If you’re building anything that depends on chip supply over the next three years, treat Intel’s revival as a hedge worth tracking, not a plan worth betting on. Specifically:
- Don’t single-source your roadmap on TSMC assumptions. Intel coming back online as a real foundry option changes pricing leverage even if it never matches TSMC’s leading-edge nodes.
- Watch the Apple and Tesla deals for substance. Preliminary agreements are cheap. Volume commitments at specific nodes are the signal.
- Separate the political story from the technical one. Government backing buys runway. It doesn’t fix yields.
Tan’s track record at Cadence earned him this shot. He turned that company around by being patient and surgical. Intel investors are pricing in the same outcome at roughly 5x the speed. The multibillion-dollar question is whether the execution catches up to the stock chart, or the stock chart eventually catches down to the execution.
Full breakdown is over at the original TechCrunch AI piece.