TER · Teradyne, Inc. · Stock research

Current Thesis

The July drawdown was a positioning event and the print settled it. Teradyne had guided Q2 revenue down sequentially to $1,150–1,250M; it delivered $1,329M on 2026-07-28, above the top of its own range and +104% year-on-year, with Semiconductor Test at $1,122M. Non-GAAP EPS was $2.47 against a $2.05 consensus (StockStory, 2026-07-28), and the shares rose 12.8% that session. Price has retraced from the 2026-07-17 close of $322.36 back to $418.79 on 2026-08-14 — roughly a 30% recovery — without reclaiming the $483.84 closing high of 2026-06-30. What has changed since the last note is the direction of the gap between price and the sell-side: consensus target across 18 analysts is now $406.65, below the last close, where in mid-July the entire $446–$550 band sat stranded above spot. The narrative leg being bought is test intensity per AI wafer — more HBM stacks, more advanced-package good-die screening, more SOC test on merchant and custom accelerators — funded by a wafer-fab-equipment cycle CEO Greg Smith described on 2026-07-28 as setting "the stage for continued growth in 2027." There is no company-specific event to resolve that until the Q3 report in late October.

Bullish and bearish views on Teradyne, Inc.

The model's bull view on Teradyne, Inc. (TER), in brief: Q2 2026 (2026-07-28): revenue $1,329M vs. a guided $1,150–1,250M, +104% YoY; Semiconductor Test $1,122M, Product Test $107M, Robotics $100M. The bear view: Sequential math is still negative on the bottom line: non-GAAP EPS fell from $2.56 (Q1) to $2.47 (Q2) on higher revenue, and the Q3 guide of $1.85–2.15 sits well below both, on revenue guided down from $1,329M. Both cases follow in full.

Bull Case

• Q2 2026 (2026-07-28): revenue $1,329M vs. a guided $1,150–1,250M, +104% YoY; Semiconductor Test $1,122M, Product Test $107M, Robotics $100M. Second consecutive record quarter, and a beat against a range management had already cut.
• Q3 2026 guidance of $1,200–1,300M revenue and $1.85–2.15 non-GAAP EPS came in roughly 21% above the analyst midpoint at the time, per StockStory (2026-07-28) — the guide, not the beat, drove the 12.8% session move.
• Gross margin 59.8% in Q2 2026, with $734.3M of operating cash flow in the first six months and $74.2M of buybacks over the same period (Q2 press release, quarter ended 2026-06-28).
• Q1 2026 (2026-04-28) disclosed the first merchant-GPU SOC-test win at ~$50M of FY2026 revenue with multi-system production orders; Q2's Semi Test line of $1,122M versus Q1's $1,111M is the first sequential datapoint on whether non-memory test scales alongside HBM.
• TSMC's 2026 capex guide of $60–64B (raised from a $56B ceiling) is the upstream funding source for test intensity; the Q2 call framed 2027 as a continuation, not a digestion year.
• Post-print target revisions skewed up: JPMorgan to $465 from $450, Evercore ISI to $420 from $370, Morgan Stanley to $397 from $387 (late July 2026).

Bear Case

• Sequential math is still negative on the bottom line: non-GAAP EPS fell from $2.56 (Q1) to $2.47 (Q2) on higher revenue, and the Q3 guide of $1.85–2.15 sits well below both, on revenue guided down from $1,329M.
• Consensus target $406.65 is beneath the 2026-08-14 close of $418.79, and Baird cut to $420 from $446 into the beat. Dispersion is extreme — the published range runs $270 to $550, with 12 buys, 6 holds, no sells.
• RSI(14) at 71.5 on 2026-08-14 with the stock 13.4% under the $483.84 high leaves a band of June supply overhead and no company event before the late-October print to absorb it.
• Advantest (6857.T) holds the incumbent HBM memory-test socket, and SK Hynix has been internalizing system-level HBM4 test (TrendForce, 2026-02-25) — direct structural competition against the growth leg the 2026 re-rate was built on.
• Concentration was flagged by management on the 2026-04-28 call: an AI mix around 70% routes revenue through few vertically integrated customers, so single-program air pockets are large relative to a $1.2–1.3B quarter.
• Cash and equivalents of $349.5M at 2026-06-28 is a thin buffer relative to the market capitalisation being paid for the growth path; the balance sheet is not the reason to own this.
• The April–June template ran twice already in 2026: record results on 2026-04-28 sold 17–19%, and the 2026-06-30 closing high preceded a 33% drawdown in twelve sessions. Price front-running the print has repeatedly been the risk.

Setup & Price Structure

The narrative is maturing. The narrative is well known, still working, and the flow is moderating rather than expanding. What dates that call: the re-rating headline is the 2026-07-28 print, now three weeks old; the recovery from $322.36 (2026-07-17) to $418.79 (2026-08-14) has been continuous but has not tagged $483.84; and the sell-side response to a 20% EPS beat was a mixed set of revisions, including one cut, leaving the 18-analyst consensus target below the last close. Fresh attention is arriving through sector headlines (2026-08-07, chip stocks bid after an AI-trade shakeout) rather than through new company disclosure.

Crowding and positioning observables, stated as observables: RSI(14) 71.5 on 2026-08-14, a three-month return of +24%, price 13.4% below the 52-week high, consensus target $406.65 sitting under spot, and CEO Gregory Smith's 2026-07-15 sale of 4,000 shares at a $356.31 weighted average under a 10b5-1 plan adopted 2026-02-12 (116,495 shares retained). The insider sale was pre-scheduled and carries little standalone signal; the plan-adoption date is what matters if further sales cluster near the highs. No new Form 4 has surfaced since.

Structure: the July flush low at the 2026-07-17 close of $322.36 is the reference floor; the post-print advance is the leg that has to hold. Half of that advance sits near $370. The $483.84 closing high from 2026-06-30 is the unfinished business, and a failure to reach it before the Q3 print would leave a lower high on the weekly chart for the first time this cycle.

Catalyst Calendar (next 30 days)

• 2026-08-26 — NVIDIA Q2 FY2027 results and call (quarter ended 2026-07-26). The largest single read-through on accelerator volumes and HBM attach; Teradyne trades as sector beta into it.
• 2026-08-27 — Marvell Technology quarterly results. Custom-silicon/ASIC volume commentary is the second-order read on SOC test demand.
• 2026-09-02 — Broadcom Q3 FY2026 results (company-confirmed date). XPU order trajectory is the cleanest public proxy for non-merchant accelerator test units.
• ~2026-09-10 (est.) — TSMC August monthly revenue release (published around the 10th each month). Monthly foundry loading is the highest-frequency check on the capex assumption behind the 2027 growth comment.
• ~2026-10-27 (est.) — Teradyne Q3 2026 report, outside the 30-day window and the next company-specific binary. It is the first test of whether the $1,200–1,300M guide was conservative.

What Would Change Our Mind

The structural break is the post-print advance failing before it reaches $483.84 and then unwinding — a lower weekly high plus loss of the recovery leg would say the 2026-07-28 beat was absorbed rather than believed. Expressed as a gradeable level: a weekly close below $370 hands back half the advance from the 2026-07-17 close of $322.36 and puts the July low back in play. Secondary conditions, each independently observable: NVIDIA on 2026-08-26 or Broadcom on 2026-09-02 guiding accelerator or capex commentary flat-to-down; a Q3 report at or under the $1,200M low end with no 2027 reiteration; Advantest raising HBM-test share guidance at its next quarterly report while Teradyne's Semi Test line goes sequentially negative; or additional insider sales clustering above the $356.31 weighted average of the July 10b5-1 execution. On the other side, a weekly close above $483.84 with Semi Test guided above $1,300M would put the theme back into an accelerating phase.

Correlation Notes

• Moves with the WFE and test complex — AMAT, LRCX, KLAC, ONTO and Advantest (6857.T) — at near-1 correlation on sector flushes; the 2026-06-05 and 2026-07-01/02 sessions moved the group together. Owning several of these is one exposure, not several.
• Directionally levered to TSMC capex guidance ($60–64B for 2026) and to monthly foundry revenue; a capex cut is transmitted to test orders with a lag of quarters but to the share price same-day.
• Memory pricing and HBM4 qualification news (NVIDIA certified Samsung, SK Hynix and Micron in early 2026) drives the memory-test leg; Advantest headlines should be tracked independently because share shift between the two is the zero-sum part of the story.
• Robotics (Universal Robots / MiR) at $100M in Q2 2026 versus $1,122M in Semi Test is roughly 7.5% of revenue — the Vention modular-automation partnership announced 2026-06-22 is optionality on the physical-AI theme, not a driver of the current multiple.