The indium phosphide boom behind AXT (NASDAQ: AXTI) is not a fantasy. That is the easy part. The harder question is whether a company that just crossed into profitability for the first time deserves a market capitalization of 5.12B after a 650% year-to-date rally. It does not. I'm maintaining a Hold on AXTI. The growth story is intact, but the multiple has outrun the proof.
What actually changed
AXTI makes compound semiconductor substrates - the starting wafers that companies like Coherent, Lumentum, and Innolight use to build optical transceivers for AI data centers. Those transceivers move data between GPUs at 400G, 800G, and now 1.6T speeds. Indium phosphide is the material of choice for the laser components inside those transceivers. The supply chain is narrow, and AXT has positioned itself as one of the few Western-accessible sources.
The demand surge is real. In Q1 2026, AXT reported revenue of $26.9 million, up from $23.0 million in Q4 2025 and $19.4 million in Q1 2025. Indium phosphide alone generated $13.6 million - roughly half the company's revenue, compared to a fraction of the total a year ago. The legacy LED and silicon substrate businesses are fading, but nobody is missing them. InP is the entire point now.
Management guided to quarterly profitability as early as Q2 2026, with expected net income of $0.06 to $0.08 per share. Wall Street consensus projects $14.6 million in annual profit for 2026. That's a narrative shift from the $1.59 million GAAP operating loss the company posted in Q1, even as non-GAAP gross margin climbed to 29.9%.
The capacity expansion plan - funded, aggressive, and dilutive
This is where the story gets more interesting, and also where the stock's current valuation runs into trouble.
AXT is doubling its indium phosphide capacity by the end of 2026 from Q4 2025 levels, and targeting a quadruple expansion by 2027. Management says it's ahead of schedule. The run rate should hit approximately 4,000 wafers per quarter by year-end. To fund this, the company closed a $632.5 million capital raise just before the Q1 earnings release. That is a large war chest for a $26.9 million-a-quarter business, and it signals conviction - but it also means the current ~$76 million diluted share count is going up. Substantially.
The balance sheet before the raise showed $128.4 million in cash and $77.3 million in debt. Add the $632.5 million in proceeds, and AXT has the capital to execute. The question is whether dilution and capex absorption keep earnings per share from meaningfully compounding, even if the top line does.
The valuation problem
Here is where the rating breaks. AXTI last traded around $91, giving it a market cap in the $5 billion to $5.5 billion range. Annualize Q1 revenue of $26.9 million and you get roughly $108 million in trailing run-rate sales. The stock trades at approximately 47x annualized revenue. It has no meaningful free cash flow yet - operating cash flow in Q1 was $4.3 million, down from the prior-year quarter, and the company is still absorbing ramp costs.
Compare that to the broader semiconductor substrate and materials space. Even high-growth, cash-flow-positive names like Entegris and Akoustis traded at far lower revenue multiples during their own expansion cycles. AXTI is being valued as if it will execute the 4x capacity ramp flawlessly, secure every incremental data center order, and hit profitability without margin compression or customer concentration risk.
The 650% YTD move tells you exactly what the market has decided. The problem is that a 650% move on a $108 million revenue company leaves almost no room for execution missteps at a $5 billion valuation. If the capacity ramp runs even one quarter behind, or if gross margins compress as volume scales through new equipment, the stock will re-rate quickly. It has already proven volatile - the $550M+ offering in late April triggered a sharp selloff even as the same offering was the proof of conviction the bulls needed.
What it would take to upgrade
I would upgrade to Buy under two conditions:
- Q2 2026 proves quarterly profitability. Management has guided to $0.06–$0.08 per share. Hitting that on a strong InP revenue print would be the first real evidence that the operating model scales. Missing it would extend the cash-burn narrative and give sellers fresh ammunition.
- The valuation resets or the growth re-accelerates further. If AXTI pulls back to the $60s on market noise - that's roughly 25x annualized revenue - the risk/reward shifts. Alternatively, if InP revenue doubles again in a single quarter to the $27M+ range, the growth rate itself could justify a higher multiple. Right now, we have neither.
The risks that determine the rating
- Customer concentration. AXT sells InP to a narrow set of optical transceiver manufacturers. If one customer shifts volume to a competitor or an internal source, revenue could drop faster than the market expects.
- Dilution. The $632.5 million raise is a positive for execution but a headwind for EPS. If the share count grows 40–50%, the earnings power per share shrinks proportionally.
- Competitive catch-up. Other InP substrate producers - particularly in China and Southeast Asia - are expanding. AXT's advantages (yield expertise, customer relationships, US supply chain positioning) are real but not permanent if pricing pressure arrives.
- Legacy business tail drag. The LED and silicon substrate segments continue to decline. They still represent roughly half of Q1 revenue and are lower-margin, which drags the blended result as the mix slowly shifts.
Investor takeaway
AXTI is not a narrative stock pretending to have a business. The indium phosphide demand from AI data center optical interconnects is genuine, the capacity plan is funded, and the path to profitability is visible. But at $5+ billion, the stock is pricing in flawless execution for two years straight. That's not a Buy - it's a wait.
Hold until Q2 earnings deliver the quarterly profit management has promised. If it does, and the stock takes a breath from these levels, the setup changes. If it doesn't, the correction will be steeper than anyone in this rally wants to admit. The next earnings report is the gate.