Elon Musk dropped an intriguing hint on Thursday: Tesla might need to build what he's calling a "gigantic chip fab" – and Intel could potentially be part of the picture.

The Intel angle

Speaking at Tesla's annual shareholder meeting, Musk casually floated the idea of working with the struggling American chipmaker. "You know, maybe we'll do something with Intel," he told the cheering crowd. "We haven't signed any deal, but it's probably worth having discussions with Intel."

It's classic Musk – part speculation, part negotiation tactic, entirely public. But the market took notice: Intel shares jumped 4% in after-hours trading on the comment alone.

Intel, for its part, declined to comment. Which is probably wise given nothing's actually signed yet.

Why this matters

Here's the context: Tesla is designing its fifth-generation AI chip to power its autonomous driving ambitions. These aren't just any chips – they're the brains behind Tesla's Full Self-Driving software and the company's broader vision of becoming an AI and robotics powerhouse.

The problem? Even with current suppliers like Taiwan's TSMC and South Korea's Samsung on board, Musk reckons it won't be enough. "Even when we extrapolate the best-case scenario for chip production from our suppliers, it's still not enough," he admitted on Thursday.

Enter the "terafab"

This is where things get interesting – or potentially concerning, depending on your perspective. Musk suggested Tesla might need to build what he's calling a "terafab." "It's like giga but way bigger," he explained with characteristic understatement.

"I can't see any other way to get to the volume of chips that we're looking for. So I think we're probably going to have to build a gigantic chip fab. It's got to be done," he said.

Now, Musk being Musk, he didn't exactly offer a detailed business plan. But he did throw out some numbers: the facility would need to produce at least 100,000 wafer starts per month. For context, a wafer start is how the semiconductor industry measures manufacturing output.

The chip specs

What would make these chips special? According to Musk, they'd be specifically optimised for Tesla's software whilst being both cheaper and more power-efficient than current options. He claimed the chip would consume roughly a third of the power used by Nvidia's flagship Blackwell chip, whilst costing just 10% as much to manufacture.

Those are bold claims, but then again, this is Musk we're talking about.

The timeline

A small production run of the AI5 chip is expected in 2026, with high-volume manufacturing only hitting its stride in 2027. Musk's already thinking ahead to AI6, which he says will use the same facilities but deliver roughly double the performance, with volume production targeted for mid-2028.

Tesla's currently on its fourth-generation chip, so this represents a fairly aggressive development roadmap.

The bigger picture

Thursday wasn't just about chips for Musk. He scored a major victory when shareholders approved a staggering $1 trillion pay package over the next decade. It's an endorsement of his vision to transform Tesla from an electric vehicle maker into what he sees as an AI and robotics juggernaut.

"I'm super hardcore on chips right now as you may be able to tell," Musk said with a laugh. "I have chips on the brain."

Intel's opportunity?

For Intel, this could represent a lifeline. The American chipmaker has been struggling, lagging far behind Nvidia in the AI chip race despite having its own manufacturing facilities. The U.S. government recently took a 10% stake in the company, and Intel desperately needs to find external customers for its newest manufacturing technology.

A partnership with Tesla – even if it's just exploratory talks at this point – could provide exactly that. But it's worth remembering: Musk floats ideas publicly all the time. Some materialise, others evaporate. Intel would be wise not to count their chickens just yet.

The reality check

Building a semiconductor fabrication plant is phenomenally expensive and complex. We're talking billions of dollars and years of development. Musk's breezy "it's got to be done" doesn't quite capture the enormity of such an undertaking.

But if there's one thing Musk has demonstrated repeatedly, it's a willingness to tackle seemingly impossible projects. Whether Tesla actually builds this "terafab" remains to be seen, but at the very least, it signals just how serious the company is about vertical integration in the AI chip space.

For now, Intel's shares are up, shareholders are happy, and Musk has chips on the brain. Whether any of this translates into actual factories and partnerships is a story that'll unfold over the coming months and years.