TER — Teradyne

Thesis

Verdict: WATCH at ~$304 (as of 2026-03-09). Medium-high conviction on the business, low on the entry price. Teradyne is the dominant Western half of a global duopoly in semiconductor automated test equipment (ATE) — it and Advantest hold ~95% combined share. AI is reshaping the demand curve: compute test revenue grew +90% YoY in 2025 and is now the largest revenue component, with AI mix heading toward ~70% of revenue in Q1 2026. The business throws off 35%+ ROIC through cycles on a clean balance sheet. The open question is whether AI-driven test demand is a structural multi-year re-rating or a cyclical peak followed by a digestion period — management itself warned of "lumpiness."

The price has already discounted much of the good news. At ~$304 the stock is up 66% YTD (2nd-best S&P 500 performer in 2026) and +362% off the 52-week low, sitting just below the all-time high of $344.92. The average analyst target (~$303) is essentially at the current price, so further upside requires multiple expansion or beats above already-elevated expectations. Valuation is stretched: trailing P/E ~77–88x, forward ~38–53x, EV/EBITDA ~48x. The single biggest source of loss is FOMO after a +66% run, not the business.

The risk/reward improves meaningfully on a pullback to $260–280, closer to the 50-day MA, where you'd own a duopoly ATE franchise at a more reasonable forward multiple. A small starter at current levels is defensible for high conviction in multi-year AI test demand; full sizing should wait for a pullback or Q1 2026 confirmation. Recommendation: WATCH — high-quality business, demanding price.

Snapshot

One-liner: Teradyne makes the machines that test every semiconductor before it ships — the final exam every AI processor, memory chip, and networking IC must pass — and is the Western leg of the Teradyne/Advantest ATE duopoly now riding an AI-test super-cycle.

• Ticker / exchange: TER on NASDAQ
• Company: Teradyne, Inc.; CEO Greg Smith (long-tenured Teradyne executive)
• Sector / industry: Semiconductors — automated test equipment (ATE); also robotics and humanoid value-chain exposure
• Spot price: ~$304 (2026-03-09) | Market cap: ~$47B
• Performance: +66% YTD 2026 (2nd-best S&P 500 performer); +362% off 52-wk low; all-time high $344.92
• Valuation: Trailing P/E ~77–88x | Forward P/E ~38–53x | EV/EBITDA ~48x | EV/FCF ~67x
• Analyst target: mean ~$303 (≈ at spot — no margin of safety); internal valuation/exit target $400+
• Conviction: Medium-high on business quality; WATCH at spot, BUY zone $260–280

Business

What they do. Teradyne makes automated test equipment — the machines that test every semiconductor chip before it ships. Think of it as the final exam every AI processor, memory chip, and networking IC must pass before it can be installed in a data center.

Revenue model. Capital-equipment sales (systems plus recurring service/support contracts). Revenue is cyclical, tied to semiconductor capex, but AI is potentially extending the cycle duration. Q4 2025 run-rate annualized to ~$4.3B against $3.19B FY2025 revenue — the exit rate is accelerating well above the trailing base.

Segment dynamics. Compute test is now the largest revenue component after +90% YoY growth in 2025; AI mix is heading toward ~70% of revenue in Q1 2026. The MultiLane Test Products joint venture extends the company's TAM from wafer-level test to data-center-level interconnect testing — a structural TAM expansion, not just more of the same.

Competitive moat.

• Duopoly structure: Teradyne + Advantest hold ~95% combined ATE market share — one of the most concentrated markets in semiconductors.
• High switching costs: test programs are written for specific platforms; re-qualifying on a competitor takes 6–12+ months.
• Vertical integration: Teradyne invested $516M in Technoprobe (probe cards), building an ecosystem around its platforms.
• R&D barrier: building ATE for 3nm/2nm chips with multi-GHz I/O is deeply specialized.

Customer concentration. Top customer is 19% of revenue; top 3 "specifiers" represent 41%. Losing a major AI chip customer would hurt, but concentration is manageable for a capital-equipment business.

Adjacent exposure — humanoid robotics. Teradyne is included in Morgan Stanley's "Humanoid 100" — a global mapping of public equities exposed to the humanoid robot value chain across Brain (semis/software), Body (industrial components), and Integrators — as a company either confirmed involved or with perceived material potential to support humanoid development. Flagged as tangential to the core ATE thesis, not a load-bearing driver.

Financials

Core metrics (FY2025):

Metric	Value	Assessment
Revenue (FY2025)	$3.19B	+13% YoY; Q4 run-rate $4.3B annualized
Revenue growth (3-yr CAGR)	~0.3%	Flat due to 2023 downcycle; exit rate accelerating
Gross margin (non-GAAP)	58.3%	Excellent — expanding
Operating margin (Q4)	27.1%	Strong operating leverage
Free cash flow	$450M	Healthy
Cash	~$322M
Total debt	~$200M	Debt/equity just 7.2%
ROIC (5-yr avg)	35.7%	Outstanding value creation through cycles

No red flags on quality. Clean balance sheet, high margins, strong ROIC through cycles. The one tension is the flat 3-year revenue CAGR (~0.3%) — a function of the 2023 downcycle — against a sharply accelerating exit rate; the bull case is that AI test demand structurally resets the base.

Capital return. The company returned $785M to shareholders in FY2025 via buybacks and dividends — 174% of FCF, signaling capital-return confidence (and drawing on the balance sheet, given the cash position).

Valuation multiples:

Metric	Value
Trailing P/E	~77–88x
Forward P/E	~38–53x
EV/EBITDA	~48x
EV/FCF	~67x

At ~48x EV/EBITDA the market is pricing in ~25–30% earnings growth for the next several years — achievable if AI mix keeps expanding, but demanding for a business with a 3-year CAGR of essentially zero. At $340+ you'd be near the all-time high with consensus already met — thin margin of safety.

Industry landscape

Semiconductor automated test equipment — a Teradyne/Advantest duopoly (~95% combined share) levered to the AI compute build-out, with a separate, tangential read-through to the humanoid-robotics value chain (Morgan Stanley "Humanoid 100"). Sector-level detail belongs on the sector page rather than here. See sector page: semiconductors (create/link if/when it exists).

Management

CEO Greg Smith is a long-tenured Teradyne executive. The FY2025 capital return of $785M (174% of FCF) via buybacks and dividends signals capital discipline and management confidence. No governance red flags surfaced in the checklist work; a deeper /mgmt-dd pass has not yet been run — section left open for the next skill.

Catalysts & risks

Catalysts — near-term:

• Q1 2026 earnings is the gate — tests whether $1.15–1.25B guidance holds and whether AI mix reaches ~70% of revenue.
• AI compute-test demand proving structural (multi-year) rather than a 2–3 quarter surge.
• MultiLane JV ramping data-center interconnect test, extending TAM beyond wafer-level.

Risks (structured):

1. Cyclicality (High) — semiconductor test is inherently lumpy; management warned a "2–3 quarter surge may lead to shorter digestion periods." If AI capex pauses, revenue drops fast.
2. Advantest share gains (Medium-High) — Advantest is winning AI/HPC compute-test share (NVIDIA reportedly uses Advantest). Share shifts in a duopoly are zero-sum.
3. Valuation risk (High) — at ~80x trailing P/E, any demand disappointment triggers a sharp drawdown; the gap between price (~$304) and the 200-day MA (~$157) is ~2x, so mean reversion would be violent.

What makes the thesis wrong: AI test demand proves to be a 2–3 quarter surge rather than a structural shift, causing revenue to revert toward $2.5–3.0B and multiples to compress.

Behavioral-trap audit: FOMO is the biggest trap (stock +66% YTD, +362% off the 52-wk low — checkbox lit). Narrative seduction and recency bias flagged but unchecked — the AI test story is backed by real numbers (+90% compute-test growth), so it is more a sustainability question than a narrative one; the risk is extrapolating Q4's blowout into a permanent trend that management itself cautioned against.

Valuation / DCF

No formal DCF modeled. Multiple-based read: expensive on trailing (~77–88x P/E, ~48x EV/EBITDA), more reasonable on forward (~38–53x P/E) if AI test demand sustains. The market is implying ~25–30% multi-year earnings growth.

Downside scenario: in a cyclical downturn TER could revert to ~$2.5B revenue with 15–20% operating margins, implying fair value around $150–180 (40–50% downside).

Entry / exit zones (the 50-day MA sits at ~$254 as of 2026-03-09 — the reference level for both the BUY zone and the mental stop below):

• WATCH / BUY $260–280 — pullback toward the 50-day MA (~$254); better forward multiple, real margin of safety vs the ~$303 consensus target.
• Valuation target / trim $400+ — internal exit/valuation target.
• Hard re-evaluate: mental stop at $225 (below the 50-day MA), where the AI narrative would be questioned; max loss tolerance 25%.

Decision log

2026-03-09 / 2026-03-14 — Pre-buy checklist: WATCH at ~$304, Medium-high conviction. Scorecard: thesis clear ✓, business understood ✓, financials healthy ✓ (58.3% non-GAAP GM, 35.7% 5-yr ROIC, D/E 7.2%), valuation No (trailing P/E ~77–88x, forward ~38–53x), behavioral trap Maybe (FOMO after +66% YTD), technicals Mixed (strong uptrend, RSI ~68–70 approaching overbought, price ~2x the 200-day MA). Average analyst target ~$303 ≈ spot, so upside needs multiple expansion or beats above elevated expectations.

• Single most important thing that must go right: AI-driven test demand proves structural (multi-year), not a 2–3 quarter surge + digestion.
• Expected holding period: 2–3 years.
• Exit criteria: sell if AI mix declines for two consecutive quarters (demand digestion), if Advantest takes dominant compute-test share, or at $400+ (valuation target).

Position plan (from checklist). Target size 3–4% of portfolio. Scale in: start with a 2% position on a pullback to $270–280; add another 1–2% on Q1 2026 earnings confirmation. Maximum loss tolerance 25% — mental stop at $225. Small starter at spot is defensible only for high conviction in multi-year AI test demand; full sizing should wait for the pullback or Q1 confirmation.

Net stance: a high-quality duopoly business riding a genuine AI tailwind, but the stock has already priced in much of the good news. WATCH; buy the pullback, not the parabola.

Sources

Fragments folded into this canonical page (consolidated 2026-06-02; original archived to _migration-archive/2026-06-02/TER/): ter-buy-checklist.md (pre-buy checklist, dated 2026-03-09, written 2026-03-14). Auto-maintained "Humanoid 100" source-update note (Morgan Stanley, Mar 19 2026 drop) folded into Business / Industry landscape.

Filings & earnings history: ter-filings (sanctioned second file — chronological filings/earnings; not yet populated, run /filings TER).

Source documents: dropfile://Robotics/The_Humanoid_100_-_Mapping_the_Humanoid_Robot_Value_Chain.pdf (Morgan Stanley Humanoid 100).

Related vault pages: semiconductors (sector, if/when created).