This is the part of the AI boom that makes me nervous: not the flashy demos, but the quiet moments when a company posts “record earnings” and everyone starts treating a single demand wave like a permanent law of nature.
Teradyne just reported record Q1 2026 results, and the headline numbers are undeniably strong. Revenue came in at $1.28 billion, ahead of the $1.21 billion estimate. Adjusted earnings were $2.56 per share versus the $2.10 people expected. That’s not a “pretty good quarter.” That’s a real beat.
And the driver is even clearer than the numbers: AI-related demand is now about 70% of its revenue. That’s the tell. Teradyne isn’t just benefiting from “tech doing well.” It’s riding a very specific wave: the world buying, building, and testing the chips that power AI.
If you want the simple version of what this means: Teradyne is one of the companies selling the picks and shovels. Its Semiconductor Test division led the quarter with $1.111 billion in revenue. When chip makers sprint, they need more test gear. When they slow down, test gear gets delayed, pushed out, “re-forecasted.” This is a business that can look smooth in a chart until it suddenly isn’t.
The bullish case is obvious. If AI keeps expanding into everything—data centers, devices, cars, factories—then testing capacity becomes a bottleneck. Teradyne becomes the toll booth. In that world, this quarter isn’t an outlier. It’s the early shape of a longer run.
But that 70% number cuts both ways, and I don’t think people sit with that enough.
When one trend becomes most of your revenue, you don’t just get growth. You get dependency. And dependency changes how you make decisions. You start staffing to match the surge. You start building plans that assume the surge continues. You start telling your board a story that only works if the same customers keep spending at the same pace. Then, if anything cracks—budgets tighten, a big buildout pauses, inventory stacks up—you don’t gently drift down. You lurch.
Imagine you’re running a data center program and you’ve been told to spend aggressively to “keep up” in AI. Six months later, priorities change: power limits, delays in getting chips, a new internal mandate to cut costs, or simply realizing you over-ordered. The easiest lever to pull is timing. You don’t cancel the future. You delay it. And companies like Teradyne feel that delay fast.
On the other hand, if you’re a chip company in an AI arms race, you might decide the opposite: you can’t afford to pause. If you stop investing and your competitor doesn’t, you risk falling behind on performance, supply, and customer trust. In that scenario, the spending becomes stubborn. Even if it’s painful, it continues. That’s the upside Teradyne investors are implicitly betting on: not just demand, but fear-driven demand.
There’s also the robotics piece, which is smaller right now—$91 million in revenue—but interesting because it hints at a second engine. The company is expanding robotics into AI applications for data centers and manufacturing. I like the direction. But I’m not ready to treat it as diversification yet. A small segment with a big story can still be a small segment. People love to talk about “the next leg of growth” right when the current leg is doing great. That’s convenient timing.
Here’s what I do respect about this quarter: it shows Teradyne is positioned where the money is actually flowing, not where people are just talking. When AI investment is real, it shows up in orders for boring, critical equipment. Testing is not optional if you want reliable chips at scale. So yes, this is a legit signal.
My issue is the mood that tends to follow results like this. Everyone starts acting like “AI demand” is a single smooth line upward, as if companies don’t panic-buy and then pause, as if budgets don’t get reallocated, as if a few big players don’t control huge chunks of the spend.
And if AI is 70% of revenue, then Teradyne’s quarter is also a story about concentration risk. Not scandal risk—just normal business risk. When your world narrows, your outcomes get louder.
So I’m left torn in a very practical way. If you’re a worker there, a supplier, an investor, even a customer building plans around this momentum, you have to ask yourself: are we watching the start of a long build cycle, or the frothiest part of a rush that will look obvious in hindsight?
How much of this AI-driven spending is truly durable demand, and how much is fear of being left behind?