$AEHR and the Silicon Photonics Burn-In Question: Real Chokepoint, Priced-In Thesis

So here's what just happened.

$AEHR +14.7% today. Another follow-on production order from its lead silicon photonics customer for a fully automated FOX-XP wafer-level burn-in system.

That's the third follow-on order from this customer in less than five months.

The headline says "silicon photonics order." The real question is: is wafer-level burn-in a chokepoint or just another test step in a crowded chain - and is the answer already priced into a stock that went from $20 to $92 in 2026?

The signal that matters

Here's the order cadence:

• March 31 - New major silicon photonics customer announced (global networking leader). Stock jumped 20%+.
• June 17 - First follow-on order from that same customer. Stock jumped 18%.
• July 9 - Second follow-on order. Stock up 13%.
• August 4 - Third follow-on order. System ships H1 2027.

Nine independent WaferPak test blades, each delivering up to 3,500 watts. Fully automated, hands-free operation with robotic wafer handling already demonstrated at the customer site.

The customer took delivery of their first system in FY2026, qualified it, proved automated operation, and is now ordering another. That's not engineering curiosity. That's capacity expansion.

And Aehr's own numbers back it up. Q4 FY2026 bookings hit a record $60.7 million. Effective backlog is $100.6 million - basically covering the entire $50 million they booked for all of FY2026. Cash on the balance sheet: $116.5 million. Fiscal 2027 revenue guidance: $130M-$150M, which is 2.6x to 3.0x last year.

The chokepoint check

Here's where I need to slow down and map the actual dependency.

Silicon photonics mass production has three testing steps. Industry sources inside OSAT circles call them Insertion 1, 2, and 3:

→ Insertion 1: Standalone photonic IC testing - relatively mature solutions exist. Advantest's Triton platform with FormFactor probe cards is in qualification.

→ Insertion 2: The "electrical-on-top, optical-on-bottom" dual-sided wafer test after TSMC vertically stacks the electronic IC on the photonic IC. The industry calls this a "black hole." No production-worthy automated solution exists yet. Advantest, Teradyne, FormFactor, MPI Corp, and Hermes Testing are all racing to solve it.

→ Insertion 3: Post-CPO-packaging final product testing - mature solutions available.

Aehr's wafer-level burn-in sits adjacent to this chain. WLBI isn't functional testing - it's reliability screening. It catches infant mortality failures before packaging by stressing the wafer under high power and temperature. For silicon photonics chips with integrated lasers, that matters because optical coupling misalignment by a few micrometers creates loss that electrical tests won't catch. A device can pass all electrical tests and still fail in the field.

WLBI is becoming a foundational manufacturing step as laser integration gets tighter and production volumes ramp. The question is: how concentrated is the supplier base?

The competition reality

This isn't a monopoly. Worth stating plainly.

Advantest and Teradyne - the two giant ATE (automated test equipment) companies - are both accelerating silicon photonics test solutions. Advantest specifically mentioned PoC for Insertion 2 in the first half of 2026. They have deep customer relationships, massive install bases, and government-backed R&D budgets. They're not sleeping on this.

Aehr's edge is that WLBI is niche enough that the giants haven't fully owned it. FOX-XP can burn in nine 300mm wafers in parallel, delivering massive power per blade, at a cost structure that works for production. That's not trivial.

But it's also not "no one else can do this." The moat here is customer qualification and switching cost, not architectural monopoly. Once a customer's line is running on FOX-XP with automated handling integrated, they don't want to rip it out. Different kind of moat. Same bottom line: it's real, just narrower than the purest bull case assumes.

Structure vs. price

This is where it gets interesting.

AEHR started 2026 near $20. It's at $91.71 now. One of only five stocks in the Russell 3000 to deliver 200%+ returns this year. Total return over the last year: 354.9%.

Market cap is above $2 billion on $50 million in FY2026 revenue. P/B is about 11.4x, versus the US semiconductor industry average of roughly 4.6x and peer average of about 4.8x.

The stock passes 1 of 6 standard valuation checks on Simply Wall St's screen. The math isn't on the bear side of "overvalued." It's on the bull side of "has to execute flawlessly."

The structural thesis is real. Follow-on orders from a major silicon photonics customer, moving from qualification to volume, is exactly what the market was waiting for. The $100M+ effective backlog de-risks FY2027. The $116.5M cash position means no dilution risk for the foreseeable future.

But the market has already rewarded the thesis. Repeatedly. Three times already on follow-on orders this year. The stock also surged nearly 26% after Q4 earnings in July - the biggest pop of the fiscal year - despite a GAAP net loss for the full year. The market is trading the bookings, not the income statement. That's correct, but it also means the price is front-running revenue recognition.

My read: the chokepoint is narrower than most people think, but it's wider than the bull case assumes. WLBI for SiPh is a real growth niche with limited competition. Aehr is winning customers fast. But Advantest and Teradyne are real threats at the adjacent dual-sided test layer, and the $2B+ market cap means execution has to be flawless and the TAM has to expand - into AI processor WLBI (they're benchmarking with a top-tier AI processor supplier right now), HBM, and NAND - to justify further rerating.

> TLDR: The silicon photonics WLBI story is moving from qualification to volume. The follow-on order cadence proves it. Structure is real. Stock is up 350% in a year. P/B is 11.4x on a sector that averages 4.6x. The next rerate needs more than one customer - it needs the broader AI processor burn-in pipeline to convert, and it needs the giants to fall short at Insertion 2. The setup is defensible. The price has already done the heavy lifting. This isn't a $20 setup anymore.