{
  "@context": "https://orbyd.app/schemas/dossier.v1.json",
  "ticker": "TER",
  "name": "Teradyne, Inc.",
  "url": "https://orbyd.app/dossiers/TER/",
  "json_url": "https://orbyd.app/dossiers/TER.json",
  "status": "DORMANT",
  "current_conviction": "LOW",
  "graded_conviction": "MEDIUM",
  "archetype": {
    "code": "a2",
    "n": 2
  },
  "current_thesis": "Parabolic AI-test leg broke hard: $483.84 closing ATH (2026-06-30) to $322.36 (2026-07-17), −33% in twelve sessions on a memory-glut scare, with the whole $446–$550 analyst band now stranded above spot. Company guided Q2 revenue DOWN sequentially to $1.15–1.25B. The 2026-07-28 print is the binary that decides shakeout vs. cycle top.",
  "invalidation_trigger": "A weekly close below $296 breaks the 2026-07-17 flush low and confirms the AI-test de-rate is a trend change rather than a shakeout; secondary break is a Q2 print (2026-07-28) landing at or under the $1.15B low end of company guidance with no second-half raise.",
  "catalyst_date": null,
  "outcome": "PLAYED_OUT",
  "outcome_date": "2026-06-15",
  "invalidation_fired": false,
  "themes": [
    "semi-foundry-equipment",
    "ai-chips-memory",
    "freight-logistics",
    "physical-ai-robotics"
  ],
  "tags": [],
  "sources": [],
  "notes": [
    "April 2026 template repeats: a record beat (Q1, 2026-04-28, +87% YoY revenue) was sold ~17-19%. Trade the post-earnings reaction bar, never the headline number. The same front-run condition existed into 2026-06-30 at $483.84 and resolved the same way, only larger.",
    "Analyst price targets are lagging here, not leading: Cantor $550 (2026-06-29), Susquehanna $550 (2026-06-30), BofA $525 (2026-06-29), Goldman $465 (2026-07-06), Baird $446 (2026-06-23), consensus $423.41 across 18 analysts — every one of them set above a spot of $322.36. A wide PT-to-spot gap after a 33% break is a stale-model artifact, not upside.",
    "Advantest (6857.T) holds the incumbent HBM memory-test socket and SK Hynix is internalizing system-level HBM4 test — the two idiosyncratic offsets to the memory-test growth leg. Monitor Advantest's news feed independently of TER.",
    "Same capex trade as AMAT/LRCX/KLAC/ONTO/Advantest — correlation near 1 on sector flushes (they moved together 2026-06-05 and again 2026-07-01/02). Stacking across the complex is concentration, not diversification.",
    "First merchant-GPU SOC-test win (disclosed 2026-04-28): ~$50M FY2026 revenue, multi-system production orders, rising midterm. Whether this scales is a specific line item to check on the 2026-07-29 call.",
    "Pre-scheduled, so low signal value on its own — but the plan-adoption date matters if further sales cluster.",
    "Robotics (Universal Robots / MiR) is the non-semi optionality: Vention modular-automation partnership announced 2026-06-22. Small revenue contribution today; treat as a call option, not a thesis leg.",
    "Structural note: 52-week range $89.18-$487.91 and +242% YTD at $322 means the entire 2026 move is still intact even after the break. Percentage drawdowns off a chart like this can run far deeper than they feel."
  ],
  "body_markdown": "## Current Thesis\nThe AI-test narrative did not break on fundamentals; it broke on positioning. Teradyne closed at a record $483.84 on 2026-06-30 after a run that had it up roughly 415% year-to-date, then lost a third of its value in twelve sessions — $322.36 on 2026-07-17. The trigger was a sector-wide memory-glut scare on 2026-07-01/02, compounded by TSMC's capex-and-margin reset and a broad de-rate that produced the worst month for chip stocks versus software on record. Nothing company-specific has been disclosed. What has changed is that every buyer above $360 is now underwater into a binary print, and the company's own Q2 guide calls for a sequential revenue decline. The 2026-07-28 report is the event that decides whether this was a shakeout inside an intact capex upcycle or the first leg of a cycle top.\n\n## Bull Case\n- Q1 2026 (reported 2026-04-28) was a record: revenue $1.282B, +87% YoY, non-GAAP EPS $2.56, Semiconductor Test above $1B for the first time at $1,111M, AI-linked demand ~70% of mix versus ~60% in Q4 2025.\n- TSMC raised 2026 capex guidance to $60–64B from a prior ceiling of $56B and lifted full-year revenue growth guidance to slightly above 40% — a direct positive read-through to test intensity that the market sold anyway.\n- The sell-side band re-rated hard into the peak and has not been cut: Cantor $550 (2026-06-29), Susquehanna $550 (2026-06-30), BofA $525 (2026-06-29), Goldman $465 (2026-07-06), Baird $446 (2026-06-23), consensus $423.41 across 18 analysts.\n- HBM4 is a genuine test-intensity step-up: NVIDIA certified Samsung, SK Hynix and Micron for HBM4 in early 2026, with Micron targeting ~15,000 HBM4 wafers/week by end-2026 (TrendForce, 2026-01).\n- A second growth leg is validating outside memory — the first merchant-GPU SOC-test win (2026-04-28) carries ~$50M of FY2026 revenue with multi-system production orders and rising midterm contribution.\n- The 2026-07-17 session traded $296.82 to $330.68 and closed $321.50, in the upper third of a 10%+ range. That is the first bar since the break where sellers lost control of the close.\n\n## Bear Case\n- Management guided Q2 revenue to $1,150–1,250M and non-GAAP EPS to $1.86–2.15 — a sequential step down from $1,282M and $2.56. An upcycle narrative colliding with a guided-down quarter is exactly the setup that punishes late longs.\n- The April template already ran once: a record beat on 2026-04-28 was sold 17–19% because price had front-run the print. The condition into 2026-06-30 was the same, only more extreme, and it resolved the same way.\n- The −33% drawdown from ATH has produced no base. Twelve sessions is not a bottom; it is a first leg. There is no higher low, no reclaim of a prior shelf, and no volume signature of accumulation.\n- Analyst targets from $446 to $550 all sit above a $322 tape. Wide PT-to-spot gaps after a violent break reflect models that have not been revised, and they invite exactly the wrong kind of buyer.\n- Advantest (6857.T) holds the incumbent HBM memory-test socket, and SK Hynix is internalizing system-level HBM4 test (TrendForce, 2026-02-25) — structural competition against the leg the whole re-rate was built on.\n- Concentration was flagged by management on the 2026-04-28 call: a ~70% AI mix routes revenue through fewer vertically integrated customers and a smaller set of large device programs, so order flow is lumpy and single-customer air pockets are large.\n- At 59.9x trailing earnings and +242% YTD, valuation offers zero cushion if the second-half order commentary softens on 2026-07-29.\n\n## Setup & Price Structure\nClosing ATH $483.84 on 2026-06-30, intraday high $487.91. The break came in stages: a −13.6% session in the first week of July took price from a $427.34 close to $369.09, a bounce to $379.52 on 2026-07-06 failed, an 8.6% sector-wide drop followed, and the low print so far is $296.82 on 2026-07-17. Spot $322.36 on 2026-07-17 sits roughly 34% below the high. The 52-week range is $89.18–$487.91, so the drawdown has retraced only a fraction of the 2026 advance — there is a great deal of air beneath current price before any 2026 support becomes meaningful. Price is decisively below its rising short-term averages and the structure is a lower-high, lower-low sequence for the first time this year. The observable that matters near term is $296.82: holding it keeps the shakeout reading alive, losing it on a weekly basis converts the break into a trend change. On the upside, the $369–$380 zone where the first bounce failed is the level a genuine recovery has to reclaim.\n\n## Catalyst Calendar (next 30 days)\n\n- **Late July 2026** — semicap peer prints (AMAT/LRCX/KLAC/ONTO cluster) and Advantest (6857.T) memory-test commentary. Peer guides will reprice TER regardless of its own numbers.\n- **Ongoing through August 2026** — memory-pricing datapoints and HBM4 wafer-allocation updates; the July selloff was a memory-glut scare, so contract-price prints are the direct disconfirming evidence.\n\n## Elapsed catalysts\n\n- **2026-07-28** — Q2 2026 results, 16:30 ET. Guide: revenue $1,150–1,250M, non-GAAP EPS $1.86–2.15. The binary. *(passed 12d ago)*\n- **2026-07-29** — Q2 conference call, 08:30 ET; presentation materials from 07:30 ET. Second-half order commentary, HBM4 test bookings, and whether the merchant-GPU win scales beyond ~$50M FY26 are the three specifics that matter more than the headline number. *(passed 11d ago)*\n\n## What Would Change Our Mind\nThe constructive case requires a base, not a bounce. A weekly close below $296 breaks the flush low and reclassifies this as a trend change — at that point the memory-glut scare has become a memory-glut fact and the whole 2026 re-rate is in question. A Q2 print landing at or under the $1.15B low end with no second-half raise does the same thing on fundamentals. Conversely, a print that holds or lifts full-year targets while confirming HBM order build, followed by a reclaim of the failed $369–$380 bounce zone on a higher low, restores the accelerating read. Absent either resolution, this is a name with a broken chart, a guided-down quarter, and an earnings event inside two weeks — the combination that argues for standing aside rather than catching it. Buying because analyst targets sit $100–$230 above spot is the specific mistake this tape is set up to punish.\n\n## Correlation Notes\nTER trades as one unit with the semicap complex — AMAT, LRCX, KLAC, ONTO and Advantest all moved together on the 2026-06-05 flush and again on 2026-07-01/02. Adding TER alongside any of them is a size increase on a single capex bet, not diversification. Second-order correlation runs through memory: Micron's contract pricing and HBM4 allocation are the upstream driver of the test-intensity story, and Micron's tape is the cleanest real-time proxy for whether the glut scare has legs. Third leg is the AI-infrastructure beta itself — the 2026-07-17 framing of chip stocks having their worst month ever versus software describes a rotation out of hardware capex into software, and TER sits at the sharp end of that rotation. The robotics segment (Universal Robots, MiR, the Vention partnership announced 2026-06-22) is uncorrelated to all of it but too small to matter to the tape.",
  "first_seen": "2026-04-19",
  "last_analyzed": "2026-07-19T11:47:31+00:00",
  "last_synthesized": "2026-07-19",
  "last_update_source": "watchlist_research",
  "license": "Content © orbyd. Cite the canonical URL."
}