Paradis
AI & Technology Investment Research
Trio-Tech's business looks misaligned with what the market needs. It sells burn-in boards that customers can design out quite easily, and Aehr Test Systems could build a competing segment fairly quickly. It is also heavily concentrated in one customer, risky for such a small company; Aehr is concentrated too, but its customers keep coming back for more. x.comÂ
Trio-Tech re-rated sharply on Q3 results that validate its burn-in and testing tailwind. Semiconductor testing revenue rose 141% to $13.1M on higher volumes, and it booked about $8M of burn-in board orders from an unnamed GPU customer. The balance sheet remains solid with unused credit lines. There is no guidance, but orders signal visibility. Closing the gap between its 2.6x valuation and Aehr's 67x requires a facility expansion that shifts mix to higher-margin burn-in and a growing backlog. x.comÂ
Aehr Test Systems' addressable market looks tiny compared with 200 other AI companies, making it unattractive. x.comÂ
Aehr Test Systems' very lumpy machine sales typically make for a poor business model. x.comÂ
Trio-Tech is a risky but compelling micro-cap sitting in the middle of a burn-in testing bottleneck. Growth is huge year over year, potential new customers include Micron, and CPU demand should soar in the inference era while Trio-Tech handles burn-in testing for AMD. Real estate is roughly 10% of its market cap, and the multiple looks ridiculous if execution continues. @JonkooTrades notes Zacks just initiated the first Wall Street coverage, at Neutral, citing AI burn-in orders. x.comÂ
Trio-Tech's bear case has been validated: growth is slowing and margins are bad. The stock fell 30% on the quarter, which @JonkooTrades, a bull who admits defeats publicly, said he could not explain. x.comÂ
Aehr Test Systems is key to AI manufacturing and can even be a bottleneck, but at almost 20x revenue through 2028 at the current price, it is stratospherically priced. The risk is giant if something goes wrong for the company or for the AI ramp, and even a slowdown could be deadly at such high valuations. Tech moats are not the same as financial moats. x.comÂ
Trio-Tech is a high-conviction, high-risk asymmetric bet on burn-in testing demand for AI chips. Q3 revenue rose 124% to $16.5 million, new orders of $5.3 million and $2.6 million cover burn-in boards for next-gen AI GPUs, and a $10 million raise funds a Malaysia facility. At about $160 million it trades far below peers, with trailing EPS now positive. Risks: gross margin fell to 16% from 27%, dilution, insider selling, a run from $2.45 to $21.38 then a sharp pullback, and thin liquidity. x.comÂ
Aehr Test Systems captures more value than Trio-Tech because it sells proprietary burn-in equipment, while Trio-Tech sells consumable boards that go into competing equipment. Aehr's FOX-XP is the only validated wafer-level burn-in solution for silicon photonics transceivers, with multiple AI optical I/O wins and a patent moat, so it captures any migration to wafer-level testing. Its Sonoma package-level system won a record $41M hyperscaler order, and H2 FY26 bookings topped $92M. x.comÂ
Trio-Tech's gross margins are deteriorating because its testing revenue carries lower margins than its legacy segments. @yianisz counters that Trio-Tech still looks obvious and its margins will recover over the next 2-3 quarters. x.com x.comÂ
Trio-Tech is an Aehr Test Systems in the making with far more room to re-rate, at a $200mn market cap versus Aehr's $2.5bn, and an easy 3-5x is expected within a year. Both test GaN and SiC power devices for hyperscaler builds, where Morgan Stanley puts the powered shell at $2-6bn per GW. Aehr burns in at the wafer level with hard-to-replicate IP; Trio-Tech does back-end burn-in and reliability testing, with revenue growth accelerating to 124%, though execution risk is higher. x.comÂ
As LLMs commoditize, chip programs will multiply, favoring vendors paid per attempt: CEVA, Aehr Test Systems, inTEST, PDF Solutions and GSI Technology. Stable, cost-driven inference makes hardwired custom chips worth building, and every program pays for IP, yield and test, success or not. CEVA earns per-unit IP royalties, Aehr and inTEST scale with SKU count, PDF Solutions owns cross-fab yield data, and GSI is a risky memory-wall option. Ongoing architecture churn would make this early. x.com … Show more
