ATEYY · Advantest Corporation · Stock research

Current Thesis

The 2026-07-29 Q1 FY2026 print resolved the question that hung over this name through July, and it resolved it against the bear case. Q1 (quarter ended 2026-06-30) sales were ¥367.5B (+39.3% YoY, +12.0% QoQ), operating income ¥190.0B (+53.3%), operating margin 51.7% (+4.7pp YoY), net income ¥174.8B (+93.8%). Management raised FY2026 (ending 2027-03) guidance to net sales ¥1,714.0B from ¥1,420.0B, operating income ¥846.0B from ¥627.5B, and net income ¥660.0B from ¥465.5B, and lifted its own calendar-2026 tester TAM to $13.0–14.5B from the $10.9–12.2B set in April. The April "H2 digestion" language was retired in the same document that raised the addressable market ~19%.

The narrative leg an investor is buying has changed shape. Through H1 2026 the story was HBM stack complexity — test items per device rising with 3D packaging. The July raise came from somewhere else: SoC test system guidance went to ¥1,243.5B from ¥999.5B (+¥244.0B) while memory test went to ¥227.0B from ¥201.0B (+¥26.0B). Roughly nine-tenths of the raise is SoC, attributed by management to AI inference ASICs, CPUs and DRAM volume. This is now an inference-volume story wearing a test-intensity label, and the two do not have the same failure modes.

Price has confirmed the fundamental leg with a lag. The ADR closed $231.08 on 2026-08-14 after touching $235.00 intraday — a new 52-week high, above the $216.24 record of 2026-06-25 cited in prior coverage. The June–July drawdown that framed the last note is closed out.

Bullish and bearish views on Advantest Corporation

The model's bull view on Advantest Corporation (ATEYY), in brief: The guide was raised by more than the prior guide's own growth rate. The bear view: The raise concentrated the dependency. ¥244.0B of the ¥294.0B revenue revision is SoC test. If the AI inference ASIC order wave is a pull-forward rather than a run-rate, the memory line (+¥26.0B) is too small to cushion it. China is 48.6% of Q1 revenue. ¥178.6B of ¥367.5B in the… Both cases follow in full.

Bull Case

• The guide was raised by more than the prior guide's own growth rate. FY2026 operating income went to ¥846.0B from ¥627.5B on 2026-07-29 — a ¥218.5B revision, against FY2025 actual operating profit of ¥499.1B reported 2026-04-27. Guidance operating margin now 49.4%.
• Order visibility extended. The 2026-07-29 presentation puts customer forecast visibility at roughly 18 months versus about 6 months historically. Longer visibility is what allows an ATE maker to commit capacity without stranding it.
• Capacity commitment pulled forward. The 5,000-unit annual production target is now expected to be exceeded by March 2027, and the 10,000-unit target originally set for late-2028/early-2029 is being pulled forward (2026-07-29).
• Share position at the high end. V93000 SoC test market share ~66% in calendar 2025 with management pointing toward 70%+ (2026-07-29). Concentration in the platform that AI accelerator and custom-ASIC programs qualify on.
• Upstream capex already stepped up before this. TSMC's 2026 capex hike to $60–64B (2026-07-14/16) and ASML's second 2026 raise to €43–45B (2026-07-15) lead tester orders by two to four quarters and predate the raised guide.

Bear Case

• The raise concentrated the dependency. ¥244.0B of the ¥294.0B revenue revision is SoC test. If the AI inference ASIC order wave is a pull-forward rather than a run-rate, the memory line (+¥26.0B) is too small to cushion it.
• China is 48.6% of Q1 revenue. ¥178.6B of ¥367.5B in the June quarter came from China, versus Taiwan ¥69.4B and the Americas ¥67.4B. Any export-control action reaching test equipment, or domestic ATE substitution, hits the largest single geography first.
• The bar is now the raised guide. ¥1,714.0B in sales and ¥846.0B in operating income are the numbers the late-October H1 print is measured against. A guide-maintain in October, after this reception, reads differently than a guide-maintain would have in April.
• FX is a live swing factor. Guidance assumes ¥152/USD and ¥174/EUR, with each 1-yen USD move worth ±¥5.3B of operating income. Yen strength subtracts from the raise mechanically.
• Sell-side average sits below the all-time high. Aggregated consensus in mid-August 2026 shows an average 12-month target of ¥35,140 (high ¥42,400, low ¥23,000) across 20 analysts, 16 buy / 4 hold / 0 sell — an average that is beneath the 2026-06-25 Tokyo record of ¥35,940. The published targets are not underwriting a large move from the record.

Setup & Price Structure

• ADR closed $231.08 on 2026-08-14 (-0.17% on the session), day range $230.04–$235.00. The $235.00 print is the 52-week high; the 52-week low is $69.87. Market cap $167.74B; trailing PE 59.28 on EPS $3.88.
• The structure is a completed breakout over the prior record: $216.24 (2026-06-25) is now the shelf beneath price, and the June–July consolidation floor near $158 flagged in prior coverage sits far below and is no longer the operative level.
• The re-rating did not happen in the print gap. The Tokyo ordinary (6857.T) closed ¥27,935 on 2026-07-30 — the first full session after the 2026-07-29 15:30 JST release — which is below its 2026-07-24 close of ¥28,945. The move to new ADR highs was built over the following two weeks.
• The narrative is accelerating, dated by the 2026-07-29 TAM revision to $13.0–14.5B and the 2026-08-14 52-week high at $235.00. The maturing label carried in the 2026-07-26 note described a tape rejecting good news; that condition ended with the July print.
• Crowding observables, stated as observables: the ADR is up 191% over the trailing twelve months to 2026-08-14; ADR volume that session was 67,435 shares, thin enough that Tokyo-session news arrives as an overnight gap; consensus is 16 buy / 0 sell of 20; and there is no company-dated catalyst inside the next 30 days, so the next ten weeks are flow and read-through only.
• Issuance observable: the board resolved on 2026-07-31 to dispose of treasury stock as restricted stock at ¥27,935, the 2026-07-30 TSE close, following a similar disposal resolved 2026-03-31 at ¥20,925.

Catalyst Calendar (next 30 days)

• ~2026-08-26 (est.) — Nvidia quarterly print. Accelerator volume and hyperscaler ASIC commentary are the closest-dated proxy for the inference demand Advantest just guided to.
• ~2026-09-10 (est.) — TSMC August monthly revenue release. Monthly foundry revenue is the highest-frequency public check on the AI wafer starts feeding tester demand.
• ~late September 2026 (est.) — Micron fiscal Q4 print, the direct HBM/memory-test read-through for the smaller half of the guidance raise.
• ~late October 2026 (est.) — Advantest H1 FY2026 results, the first measurement against the ¥1,714.0B / ¥846.0B guide. This falls outside the 30-day window; there is no Advantest-dated event inside it.

What Would Change Our Mind

The structural break is a failed breakout: price returning inside the June–July range it just escaped. Concretely, a weekly close below $216 puts the ADR back beneath the 2026-06-25 record it cleared on 2026-08-14 and turns the post-print move into a two-week overshoot. Below that, a weekly close under $158 would take back the entire consolidation shelf identified in the 2026-07-26 note.

On fundamentals, three observables would break the leg independent of price: an unchanged FY2026 guide at the late-October H1 print, which would imply H2 tracking below the ramp embedded in ¥1,714.0B; the China revenue share falling sharply from Q1's 48.6% on export-control or substitution grounds; and USDJPY sustained below the ¥152 guidance assumption, which subtracts ¥5.3B of operating income per yen. A second consecutive TAM raise in October, or SoC test bookings holding the Q1 pace, would extend the leg instead.

The theme would flip toward saturated if the next set of upstream datapoints — Nvidia's print, TSMC's monthlies — comes in strong and the ADR fails to hold above the breakout shelf anyway. That is the exact pattern that dated the July maturing call, and it is the cleanest early signal available before late October.

Correlation Notes

• Direct read-through pair: Teradyne (TER), the other listed ATE pure-play. Its Q3 print lands ahead of Advantest's H1 and carries the same SoC-versus-memory mix question.
• Upstream: TSMC and ASML capex, which lead tester orders by two to four quarters. Both raised in mid-July 2026 (TSMC to $60–64B, ASML to €43–45B).
• Demand side: Nvidia, Broadcom and Marvell custom-ASIC programs. The July guidance raise is levered to inference ASIC volume specifically, so a slip in a single large custom-silicon program is a bigger swing factor than a slip in HBM.
• Macro overlay: USDJPY. The ADR carries the currency move on top of the Tokyo move, and the guide's ¥152/USD assumption makes the yen a translation and an earnings variable at once.
• Memory complex: Micron, SK hynix and Samsung HBM4 qualification timing drives the ¥227.0B memory test line — the smaller of the two guidance components after 2026-07-29.