Most investors meet AXT only after they start looking beneath the optical transceiver.
They see the laser. Then they see the material beneath the laser.
That material matters.
AXT supplies compound-semiconductor substrates, including indium phosphide, gallium arsenide, and germanium. Indium phosphide is especially important for the lasers and photodetectors that move data through high-speed optical networks.
So the easy thesis is that AI needs more optics, optics need more indium phosphide, and AXT owns the bottleneck.
The harder thesis is the one that matters to shareholders.
My view is that AXT has moved from speculative supplier to strategically relevant bottleneck, but at $75.90 the stock already assumes a dramatic increase in revenue, durable margins, and much cleaner execution than the company has historically delivered.
The scarcity is real.
The valuation is not waiting for proof.
What AXT actually controls
AXT does not manufacture the complete transceiver. It makes the crystalline substrates on which compound-semiconductor devices are built.
Think of the optical supply chain in layers.
Marvell can design the digital signal processor. Lumentum or Coherent can build optical components. Applied Optoelectronics can assemble transceivers. But the laser still needs a specialized semiconductor material beneath it.
That is where AXT sits.
The company's advantage is not simply a recipe. It has manufacturing capabilities in China and ownership interests in raw-material suppliers that feed its substrate business. That vertical access can matter when demand rises faster than qualified supply.
Why does that matter? Because customers cannot replace a critical substrate overnight. Materials must meet demanding specifications, production yields must be consistent, and new sources must be qualified.
AXT is selling a small but essential part of the system.
Small content can still create large leverage when it becomes the constraint.
The business has inflected sharply
AXT reported second-quarter 2026 revenue of $47.6 million, up from $26.9 million in the first quarter and $18.0 million a year earlier.
GAAP gross margin reached 44.9%, compared with 29.6% in the prior quarter and only 8.0% a year earlier. GAAP net income attributable to AXT was $11.1 million, or $0.17 per diluted share. Non-GAAP diluted EPS was $0.19.
That is not a small improvement.
Revenue more than doubled year over year while gross margin expanded by nearly 37 percentage points.
Why does that matter? Because it shows what can happen when scarce capacity meets stronger utilization and pricing. A substrate manufacturer with low factory utilization can look terrible. The same assets can become highly profitable when orders fill the lines.
But one strong quarter does not establish normalized earnings.
AXT's second-quarter revenue annualizes to roughly $190 million. The stock is pricing a business many times larger than that.
The balance sheet is strong because the share count changed
AXT ended June with approximately $412 million of cash, $33 million of restricted cash, and $304 million of short-term and long-term investments. Total liquidity was roughly $749 million, against about $84 million of short-term bank loans.
That is substantial financial capacity for a company of AXT's current operating size.
It did not come from years of retained free cash flow.
The second-quarter 10-Q shows approximately $602 million of first-half proceeds from common-stock issuance. Shares outstanding increased from roughly 55.3 million at the end of 2025 to 65.6 million at June, while first-half operating cash flow was slightly negative.
Why does that matter? AXT has reduced financing risk, but existing shareholders paid for that reduction through dilution.
This can still be intelligent capital allocation. If management uses the cash to build high-return capacity and captures the indium-phosphide opportunity, the new capital can create more value than it diluted.
But cash is not free money, and a larger company is not automatically a more valuable company per share.
Geography is both an advantage and a risk
AXT's manufacturing base and raw-material ecosystem are concentrated in China. China represented 64% of first-half revenue, while Europe represented 20%.
This structure helps the company coordinate production and access upstream materials.
It also creates a geopolitical pressure point.
China introduced export-permit requirements for indium-phosphide substrates in 2025. AXT has received permits for some destinations, including Europe and Japan, but timing for US shipments has been less certain.
Why does that matter? AXT can have demand, capacity, and qualified products and still lose revenue timing to an administrative decision it does not control.
That risk deserves a higher discount rate than the balance sheet alone would suggest.
Business quality is not stock valuation
AXT closed at $75.90 on September 24, 2026.
That price is roughly 26 times the company's annualized second-quarter GAAP EPS. That comparison actually understates the valuation problem because one unusually strong quarter may not represent sustainable earnings.
Another way to frame it is through the business the current price requires.
Using 70 million future diluted shares, a 25% net margin, a 24 times terminal earnings multiple, and an 11.4% required return, today's price implies approximately $1.4 billion of 2030 revenue.
That is more than seven times the annualized second-quarter revenue rate.
Could AI optics become large enough to support it? Yes.
Has AXT proved it can manufacture, ship, and collect on that scale? No.
What Wall Street currently expects
Five analysts tracked by S&P Global have a 12-month target range of $55 to $125, with a $91.60 average.
The low target is already below the share price. The high target assumes substantial additional upside.
That spread tells us something useful: even bullish analysts disagree sharply about how much of this inflection is durable.
I do not use their targets as valuation inputs. A 12-month trading target is not the same thing as a long-term estimate of per-share business value.
How I value AXT
I use 2030 revenue, normalized net margin, a terminal earnings multiple, and a fully diluted share count. I discount the result back 4.25 years at 11.4%.
The required return begins with the 5.11% 10-year Treasury yield and the 4.14% implied US equity-risk premium, then reflects semiconductor cyclicality, AXT's size, customer and geographic concentration, and export-control risk.
Conservative outcome
Revenue reaches $500 million in 2030, net margin averages 18%, diluted shares rise to 72 million, and the market pays 18 times earnings.
That produces a future value near $22.50 and a present value of roughly $14.
This is not a collapse. Revenue still grows far beyond the current run rate and the company remains profitable.
The poor stock outcome comes from a simple fact: growth can be real and still fall short of the growth embedded in the price.
Base outcome
Revenue reaches $1.0 billion, net margin reaches 25%, diluted shares settle near 70 million, and the stock receives a 24 times multiple.
That produces a future value near $86 and a present value of roughly $54.
This would be an outstanding business transformation.
It would still not justify today's share price under my required return.
Upside outcome
Revenue reaches $1.5 billion, net margin reaches 30%, diluted shares fall toward 68 million, and the market pays 28 times earnings.
That produces a future value near $185 and a present value of roughly $117.
This is the outcome the current stock needs investors to believe is increasingly probable.
The true downside stress test
Now assume optical demand pauses after the capacity build, gross margin normalizes sharply, export permits restrict the most valuable destinations, and management consumes part of the new cash before earning an adequate return.
In that environment, AXT could trade closer to the value of its remaining liquidity and modest earning power. My stress range is roughly $6 to $12 per share.
Do not treat current cash as a permanent floor. Cash can fund inventory, factories, acquisitions, losses, or working capital. It protects survival better than it protects a specific stock price.
The strongest opposing view
The best bullish argument is that my scenarios underestimate how constrained indium-phosphide supply will become.
If 800G, 1.6T, co-packaged optics, and future optical I/O all grow together, substrate demand could outrun qualified supply for years. AXT's upstream relationships and manufacturing knowledge would then be much harder to replicate than a spreadsheet suggests.
That argument is credible.
But scarcity attracts capacity. Customers seek second sources. Governments intervene. High margins encourage competitors. And a supplier concentrated in China must navigate risks that a demand forecast cannot solve.
The question is not whether AXT matters.
The question is how much scarcity rent survives by the time all that new revenue arrives.
What to watch out for
Watch indium-phosphide revenue separately from the broader substrate portfolio.
Watch whether gross margin remains above 40% after utilization normalizes.
Watch export-permit approvals by destination, especially shipments to US customers.
Watch inventory and receivables. Revenue quality matters more when growth is this sudden.
Watch how management deploys the newly raised capital and whether operating cash flow turns consistently positive.
Watch diluted shares, not merely basic shares.
Also watch whether customers qualify additional substrate suppliers. A bottleneck is valuable only while it remains difficult to bypass.
What could change the thesis
The thesis would strengthen at today's price if AXT demonstrated a credible path beyond $1.4 billion of revenue with 25% or better net margins, while converting earnings into cash and holding the share count near current levels.
The thesis would also strengthen if the stock fell far enough for my base case to offer a genuine margin of safety.
The thesis would weaken if gross margin fell below 30%, export approvals remained unpredictable, working capital absorbed the reported profit, or another major equity raise appeared before the existing cash produced returns.
Bottom line
AXT controls a real bottleneck in the AI optical chain.
That does not mean shareholders control all the economics of the bottleneck.
At $75.90, I estimate roughly $14 in a conservative case, $54 in a strong base case, and $117 in an upside case. The business can succeed spectacularly while the stock merely grows into its current expectations.
The practical principle is simple: when a tiny component becomes indispensable, value the scarcity - then ask how much of it the stock has already sold you.
Market prices and analyst estimates are as of September 24, 2026. Scenario values are rounded and are not personal financial advice.