As 2025 draws to a close and investors look toward 2026, the AI infrastructure market is on a steep growth curve that shows no sign of flattening out. Industry forecasts peg global AI infrastructure spending at roughly $394 billion by 2030, expanding at a compound annual growth rate near 19.4% as hyperscalers and enterprises race to deploy ever more powerful chips. That kind of sustained buildout creates structural demand not just for the chips themselves, but for every link in the supply chain that keeps them working reliably at scale.
Teradyne (TER) sits squarely in that chain, supplying the semiconductor test equipment needed to validate complex AI accelerators before they ship. Stifel recently upgraded shares to “Buy,” arguing that Teradyne is set to benefit from growing AI test revenue in 2026 as next‑gen devices push test intensity higher. The question now is whether this “picks and shovels” name still has room to run, or if the Street has already priced in the opportunity. Let’s find out.
Teradyne is a Massachusetts‑based provider of semiconductor test equipment and industrial robots. Price performance has been strong, with the stock up 57% YTD and 69% over the past 52 weeks.
At a roughly $30.5 billion market cap, shares trade at 60.9x trailing earnings versus a sector median 23.91x, and 1.98x PEG versus 1.7x. Crucially, these metrics leave little room for disappointment, so sustaining upside in 2025 and 2026 becomes important.
Teradyne’s growth story in 2026 is being shaped by where it sits in the AI manufacturing ecosystem. The company has become deeply embedded in advanced chip production, earning recognition from Taiwan Semiconductor Manufacturing Company’s (TSM) Open Innovation Platform for its work on 3D‑fabric and advanced‑node testing.
Another potential catalyst sits in the GPU and accelerator supply chain. UBS analysts now flag Teradyne as a serious contender to become a second‑source test supplier for Nvidia’s (NVDA) next‑generation Blackwell chips as Nvidia works to diversify its test and manufacturing footprint. Even a modest allocation of Blackwell‑related test volumes could move the needle because these AI accelerators are more complex and take longer to test than prior generations.
Lire la suite
