Aehr Test Systems' Q1 2026 Earnings Call: Contradictions Emerge on AI Processor Market Potential, Customer Engagements, and Flash Memory Outlook

The above is the analysis of the conflicting points in this earnings call

Date of Call: October 6, 2025

Financials Results

• Revenue: $11.0M, down $2.1M from $13.1M in the prior year (~16% YOY decline)
• EPS: $0.01 per share (non-GAAP), down from $0.07 in the prior year
• Gross Margin: 37.5% (non-GAAP), compared to 54.7% in the prior year

Business Commentary:

• Revenue Performance and AI Demand:
• Aehr Test Systems reported first quarter fiscal 2026 revenue of $11 million, although it was a $2.1 million decrease from $13.1 million in the same period last year.

• The decline in revenue was partially due to lower sales volume and a less favorable product mix, with contactor revenues contributing only $2.6 million, which made up 24% of the total revenue, a significant drop from $12.1 million, or 92%, the previous year.

• Financial Performance and Operational Expenses:

• Aehr's non-GAAP gross margin for the first quarter was 37.5%, down from 54.7% year-over-year.
• The decrease in gross margin was due to lower sales volume and a product mix that included lower-margin probers and an automated aligner.

• Non-GAAP operating expenses increased to $5.9 million, an 8% rise from $5.5 million in Q1 last year, primarily due to higher development expenses.

• AI Processor Burn-in and Testing:

• Aehr's Sonoma ultra-high-power packaged part burn-in systems and consumables saw significant growth, driven by the qualification and production burn-in of AI processors.

• This growth is attributed to hyperscalers expanding capacity and introducing new AI processors, with one significant customer placing multiple follow-on volume production orders.

• Silicon Photonics Market Expansion:

• Aehr experienced ongoing growth in the silicon photonics market, with an upgrade of a major customer's FoxXP systems to a higher-power configuration, doubling device test parallelism.
• The expansion is driven by the adoption of optical chip-to-chip communication and optical network switching, which are increasing demand for Aehr's testing solutions.

Sentiment Analysis:

• Results beat analyst expectations, but revenue declined YOY and gross margin fell to 37.5% from 54.7%. Management cited strong AI-driven engagement and orders but withheld formal guidance due to tariff-related uncertainty.

Q&A:

• Question from Christian Schwab (Not specified): When will bookings materially improve to drive revenue?
  Response: Expect additional capacity orders from the first AI wafer‑level burn‑in customer this year; timing uncertain and likely second half; orders will be announced as received.

• Question from Christian Schwab (Not specified): How many AI customers do you expect to be shipping to by fiscal year-end?
  Response: Targets are set for multiple additional AI customers across packaged and wafer-level; timing cautious, but management expects to capture several.

• Question from Christian Schwab (Not specified): Will material AI orders land this fiscal year or later?
  Response: Confident orders will come; one evaluation aligns with volume production in 2H calendar 2026, implying tool orders before then; large production orders may arrive as evaluations conclude.

• Question from Mark Schutter (Not specified): Do AI qualifications require a new product cycle and how risk‑averse are customers?
  Response: No new core product needed; adapt wafer pack pitch as needed; OSAT installs de‑risk adoption; buying typically aligns with new product transitions and capacity needs.

• Question from Mark Schutter (Not specified): Why start with Sonoma (packaged) versus wafer‑level burn‑in?
  Response: Aehr is neutral; customers use Sonoma for qual/production now and migrate to wafer‑level to avoid costly yield loss; Aehr is first with production wafer‑level solution and can quickly assess feasibility.

• Question from Bradford Ferguson (Not specified): What is the cost of waiting to system‑level burn‑in?
  Response: Rack‑level burn‑in is inefficient and power‑intensive; wafer‑level enables higher temperature/voltage stress, faster screening, and far lower power/cost, improving yield economics.

• Question from Bradford Ferguson (Not specified): Is your memory focus HBM or high‑bandwidth flash (HBF)?
  Response: Initial engagements skew to HBF; HBM also needs burn‑in; industry is scrambling to fix reliability, and Aehr’s high‑power wafer capabilities are differentiated.

• Question from Bradford Ferguson (Not specified): Are SiC makers at risk if they skip burn‑in?
  Response: Yes; data shows SiC devices fail without burn‑in; OEMs enforce requirements; Aehr’s wafer‑level burn‑in cost is pennies per die, and rigorous adopters like onsemi have gained share.

• Question from Larry Chlebina (Not specified): Does AMD’s news with OpenAI accelerate your second-processor evaluation?
  Response: Cannot comment on specifics; Aehr is engaged with all major suppliers, and broader processor momentum supports demand for its solutions.

• Question from Larry Chlebina (Not specified): For optical I/O, are you selling new machines or upgrades?
  Response: Both; more upgrades and new systems, including integrated automation for hands‑free operation.

• Question from Larry Chlebina (Not specified): Is the HBF opportunity with a new customer?
  Response: Same company with evolving requirements; new configuration is a superset and legacy tools won’t work, which could benefit Aehr.