Teradyne Says the Test Market Is Now Outgrowing Chip Equipment Spending, Sees ATE TAM Reaching $20 Billion by Decade's End
Q2 2026 earnings call, July 29, 2026
Teradyne delivered its second consecutive quarter of record revenue, with total company sales topping $1.3 billion, up over 100% year-over-year, and non-GAAP EPS of $2.47, up more than 300%. But the headline for investors isn't the quarter itself, it's a structural shift the company says it has now confirmed in the data: for the first time in over a decade, the automated test equipment market is growing faster than the wafer fab equipment market that historically dwarfed it. CEO Greg Smith told analysts that test spending as a share of total semiconductor capital expenditure has jumped from roughly 4% in 2023 to 7% in 2025 to 8% in the first five months of 2026, a trajectory management is still trying to pin down. "We're not sure whether it's going to settle... we think it's going to settle somewhere in the 7% to 9% range, but it's not going to continue to go up from there," Smith said in response to a question from UBS analyst Timothy Arcuri, who had pressed on whether the historical 8% correlation to WFE still held at higher spending levels.
The practical upshot is a meaningfully higher long-term target. With WFE capex now forecast to approach $250 billion by the end of the decade, Teradyne says the ATE total addressable market can reach or exceed $20 billion, a figure that sits above what some sell-side models had been using and above the range implied by rival Advantest's own TAM update. Smith noted his team was "surprised" Advantest hadn't already revised its TAM higher given the same demand signals. The company plans to formally refresh its long-term target earnings model on the Q4 call, a data point investors should watch closely given management's own admission that "increases in WFE spend will be a primary driver to ATE TAM growth."
Dual-Vendor Qualification Mechanics, and Where Teradyne Sits Today
One of the more granular disclosures on the call concerned how hyperscaler dual-sourcing dynamics actually play out, information that directly bears on Teradyne's ability to take share in compute test, its largest and most contested end market. Smith laid out a four-phase qualification process, opportunity to compete, developing a working solution, correlation, and production ramp, that takes 9 to 12 months from start to ramp. After the initial dual-sourced part ramps, customers enter what he called a "fast follower" phase in which incremental parts convert to the new supplier, with share climbing from 0% to roughly 30% during that window. It can then take several more years to reach a "mature dual vendor" stage where either platform can be used for a part's initial release. Right now, Teradyne has one compute customer in the mature dual-vendor phase, one in fast follower, and one that just completed correlation and is prior to ramp. That is meaningfully more concrete disclosure than the company has previously given, and it frames 2027 as the year these qualifications should start translating into visible share gains, a point Smith made explicitly: "I would say you'll start to see some of the effect of share change in 2027, but it's going to be gradual. It's a socket-by-socket thing."
Memory Strength Broadens Beyond HBM
Memory test revenue hit $212 million, a third consecutive quarter above $200 million, with book-to-bill above 2. What stood out wasn't HBM, which management had expected to be strong, but a bigger-than-anticipated DDR resurgence tied to agentic AI workloads and LPDDR demand from SOCAMM designs, alongside the first real signs of a NAND capacity inflection. "The DDR market is stronger now than we thought it would be in January," Smith said, adding that on NAND, "what's different now is we actually have seen the beginning of that inflection and are actually getting pull from our customers for more NAND capacity." Management does not expect memory to be lumpy heading into 2027, characterizing the capacity additions as durable rather than a pull-forward. The 2026 memory TAM is now expected to be more than 40% larger than 2025. The read-through is a business that is diversifying its AI exposure beyond the HBM trade that dominated 2024, though CFO Michelle Turner was candid that memory remains a drag on blended gross margin and will continue to be one into 2027.
Networking and CPO: Sizing an Early-Stage Market Honestly
Teradyne reiterated that co-packaged optics represents a $300 million to $700 million incremental market by 2028, a wide range that Smith did not try to narrow artificially. "That's our attempt to be as honest as we can be... it really comes down to how quickly the initial higher volume ramps of CPO primarily for scale-out achieve success over the 2027 time frame," he said, adding that next year's contribution is likely to land closer to $200 million, near the low end of the eventual range. This follows the company's acquisition of Quantifi Photonics and its Multilane Test Products joint venture for copper interconnect, both explicitly framed as hedges against uncertain timing of transitions from cable to backplane and from pluggable optics to CPO. Smith was careful to note that switch silicon itself, independent of the physical layer technology that wins, remains a strong and growing position for Teradyne.
Compute Test: Merchant GPU Ramp Underway, x86 Still a Gap
The company shipped its first merchant GPU order in the quarter and completed correlation with a second hyperscaler customer, pushing compute revenue up nearly 600% year-over-year within the SOC segment, though management flagged that this growth was front-loaded into the first half, with the next demand surge expected in the first half of 2027. Bank of America's Vivek Arya pushed Smith on Teradyne's historical skew toward ARM-based server CPUs versus x86, to which Smith acknowledged the company is still working to close the gap in x86 but would benefit disproportionately if the server CPU mix continues shifting toward ARM architectures.
Margins, Capital Return, and Guidance
Gross margin came in at 59.8%, up 250 basis points year-over-year but down 110 basis points sequentially due to one-time benefits that did not repeat. Third-quarter guidance calls for revenue of $1.2 billion to $1.3 billion and non-GAAP EPS of $1.85 to $2.15, with gross margin guided to 58% to 59% as mix shifts toward lower-margin memory, IST, and Robotics in the back half. Free cash flow was $378 million in the quarter and $579 million for the first half, up 150% year-over-year, giving the company room to keep funding R&D and go-to-market investment ahead of what it expects to be another year of "healthy growth" in 2027. The company bought back $69 million in shares and paid $20 million in dividends during the quarter, while maintaining a stated preference for reinvestment and accretive M&A over aggressive capital return.
Robotics and Product Test, historically overlooked next to the semiconductor test business, both grew double digits, with Robotics revenue up 33% year-over-year to $100 million as electronics manufacturing and semiconductor customers became the segment's largest end market. Management framed this simply as a lower-multiple beneficiary of the same AI data center build-out driving the rest of the portfolio, with the new production board test platform, Omnyx, and the Multilane JV both already shipping and expected to grow further in the second half.