Aehr Test Systems Touts AI Burn-In Edge, Sees H2 Bookings Jump to $60M–$80M

Key Points

• Wafer-level burn-in is Aehr's key differentiator: the company says its FOX systems and patented WaferPak enable screening of failing devices before advanced packaging and are offered as a turnkey equipment, software, automation and consumables solution for AI, data center and other advanced semiconductors.
• Management expects bookings to jump from about $20 million in H1 to between $60M–$80M in H2, with H2 revenue forecasted at roughly $25M–$30M as bookings ramp ahead of revenue recognition.
• Aehr reports production capacity of “upwards of 20” wafer-level and 20 package-level burn-in systems per month, cites an ASP of about $5M–$6M per wafer-level system, and says system margins are strong (around 60%+), indicating meaningful upside if demand continues.

Aehr Test Systems NASDAQ: AEHR used a recent presentation to outline its position in semiconductor reliability testing, with a particular focus on burn-in systems used in data center and AI processors. Company representatives said reliability, security and safety requirements in advanced computing are driving demand for both package-level burn-in and the company’s wafer-level burn-in offerings.

Focus on reliability testing and wafer-level burn-in

Aehr said it has operated in semiconductor test for nearly 50 years and primarily provides systems used for reliability or burn-in testing. Management described burn-in as part of a broader processor test flow that can include wafer test, packaging, final package test, burn-in, and additional testing steps.

The company emphasized its FOX wafer-level burn-in product as a differentiator, describing wafer-level burn-in as difficult to execute and stating it is “really the only one” doing it. Aehr said moving burn-in to wafer level allows customers to screen out failing devices before they are inserted into advanced packages, where a late-stage failure can result in scrapping the package and associated components.

Manufacturing footprint and stated capacity

Aehr said it is based in Fremont, California, where it performs final assembly and quality testing, while using contract manufacturing globally. Management said the Bay Area footprint has been helpful for engaging customers, particularly because many data center and AI-related decisions are made on the U.S. West Coast.

On production capacity, the company said it can ship “upwards of 20 systems a month” of wafer-level burn-in systems and another 20 package-level burn-in systems per month. Management characterized this capacity as well above current revenue levels and beyond the forecasts it has provided publicly.

Systems, automation, and the “WaferPak” model

Aehr described wafer-level systems that can test 20 wafers at a time, including the ability to run different wafer types simultaneously. Management said this parallelism helps reduce the cost of multi-hour burn-in cycles, which can run up to 24 hours. The company also highlighted a move toward full automation, noting that its lead AI wafer-level burn-in customer started without automation and is now shifting to a fully automated configuration.

The company discussed a “WaferPak” carrier used in its wafer-level burn-in process, describing it as a patented, portable carrier assembly that includes a thin chuck and probe-card interface assembled via a vacuum process. Management said WaferPaks are specific to a given wafer design and customers would need to purchase new WaferPaks when designs change. While sometimes characterized as consumables, the company said WaferPaks have long lifetimes (several hundred thousand insertions) and are replaced over years of use.

Management said the company’s offering is a turnkey solution including equipment, software, automation, consumables, and services. It also noted a broad patent portfolio across the U.S. and multiple international jurisdictions related to wafer-level burn-in.

Bookings outlook and Sonoma platform discussion

Aehr pointed to what it described as major recent news tied to its Sonoma package-level burn-in platform. Management said the company acquired Incal last year, which had been supplying high-temperature operating life testing tools to “a bunch of the AI guys.” Aehr said it could scale production more quickly than Incal, which it said could build about two systems per month, compared with Aehr’s stated capacity of up to 20 systems per month.

Management said it has received multiple orders following the acquisition and has recently received “a very large forecast” expected to drive business in its fiscal second half (which it noted ends in May). The company said it expects bookings to rise from about $20 million in the first half to between $60 million and $80 million in the second half. Aehr also said it expects second-half revenue of about $25 million to $30 million, following roughly “20-ish” in the first half, describing a setup where bookings are expected to ramp ahead of revenue recognition.

Aehr said expected demand drivers include the Sonoma system as well as wafer-level burn-in systems for AI, silicon photonics, gallium nitride power semiconductors, silicon carbide, hard disk drives, and additional opportunities in memory.

Customer activity, target markets, and margins

Management said it is engaged in a wafer-level burn-in benchmark with a “leading AI accelerator company,” and discussed an example production scenario of roughly 20 systems. It cited an average selling price (ASP) of approximately $5 million to $6 million per wafer-level system and said this type of AI adoption can create “hockey stick” revenue implications.

On market positioning, the company said its installed base includes many major semiconductor segments, including AI processors, automotive, data center, and silicon photonics. It also said its installed base is growing this year.

The company described its approach to burn-in as aimed at reducing “infant mortality” failures while avoiding prematurely wearing out devices. In Q&A, management discussed failure-rate sensitivity across markets, saying silicon carbide and GaN power semiconductors have been “north of 1%,” while AI is “less than 1%,” and offered an example of catching around 0.5% through burn-in to prevent first-year failures.

On profitability, management said it earns “very good margins” on package-level systems and lower margins on package-level consumables (saying material margins are roughly 30% for those consumables). It said system margins are “more like 60% plus,” wafer-level systems are higher than that, and wafer packs have the highest margins.

In discussing AI spending, management said it monitors volumes and views reliability needs as particularly important in automotive and robotics, where it expects burn-in to be mandatory due to safety implications. It said it is less focused on commoditized markets such as standard CPUs and mobile devices, though it noted that “CPUs in AI are definitely targets.”

Aehr also addressed recent operating context, saying it had pulled guidance in April 2025 amid supply disruptions and later reinstated guidance after conditions stabilized, while noting ongoing uncertainty around tariffs.

About Aehr Test Systems NASDAQ: AEHR

Aehr Test Systems develops, manufactures and sells semiconductor test and burn-in equipment used by device manufacturers to ensure quality and reliability of integrated circuits. Its products are designed for wafer-level reliability assessment, functional test and stress screening of memory devices, system-on-chips, optical components and power semiconductors. By focusing on wafer-level burn-in and testing processes, the company helps reduce cost and improve yield for high-volume semiconductor production.

The company's product portfolio includes FOX series wafer probe test and burn-in systems as well as ABTS burn-in ovens.

This instant news alert was generated by narrative science technology and financial data from MarketBeat in order to provide readers with the fastest reporting and unbiased coverage. Please send any questions or comments about this story to contact@marketbeat.com.