Google's Intel Deal Says More About Intel Than It Does About AI

What does this mean for the robotics supply chain?

Here's where I start paying closer attention. The AI chip shortage has been a headache for everyone in automation. I called my old colleague at Siemens last month, and he's still complaining about lead times for certain components. When the hyperscalers (Google, Amazon, Microsoft) are gobbling up chip capacity, it affects everyone downstream.

If Intel can actually deliver on foundry work at scale, that's potentially good news for the rest of us. More manufacturing capacity, more competition, maybe eventually better availability for the chips that go into industrial systems. That's the optimistic read.

The pessimistic read? Google's just locking up capacity that might otherwise go to other customers. We don't know yet. The details are too thin.

Should we trust these numbers?

I'll be direct: two unnamed sources and no official confirmation from either company. That's what we're working with. Bloomberg's reporting is solid, but this is basically a rumour with a dollar sign attached.

Three million chips sounds precise, but we don't know what kind of chips. We don't know the yield assumptions. We don't know if this is a firm order or something more like a capacity reservation. These details matter enormously and we simply don't have them.

When I read coverage like this, I always think about the announcements I saw from the inside during my Kuka years. The press release version and the reality were often, let's say, loosely related. Not because anyone was lying, but because deals are complicated and journalists have word counts.

What does this tell us about 2028?

Two years is a long time in this industry. When I started at Kuka in 2003, the iPhone didn't exist. Two years later, it still didn't exist. My point is that predicting technology timelines is a mug's game.

But the fact that Google is placing orders now for 2028 delivery suggests they're expecting to need a lot of AI compute in three years. That's not surprising. Everyone expects that. The question is whether Intel can actually execute.

Intel's had a rough decade. They missed mobile. They were late to GPUs for AI. Their process technology fell behind TSMC. Pat Gelsinger's turnaround plan has been, well, a plan. This kind of deal could be evidence that it's working. Or it could be Google getting a good price from a desperate supplier.

I genuinely don't know which it is. Anyone who tells you they're certain is selling something.

The bottom line

This is a supply chain story dressed up as an AI story. Google needs chips. Intel wants to make chips for other people. They apparently struck a deal. The AI angle is real but secondary.

For those of us in industrial automation, the relevant question is whether this signals a healthier, more competitive chip manufacturing ecosystem in a few years. I'm cautiously optimistic, but I've been burned by cautious optimism before (remember when collaborative robots were going to take over every factory floor by 2020?).

I'll be watching Intel's foundry numbers more closely now. That's probably the right takeaway here. Not "Google is doing something exciting with AI," but "Intel might actually be getting its act together." We'll see.

Aisha Patel · Yesterday · 5 min