This Obscure AI Stock Is Up by Almost 6,000% Over the Past Year. Can It Go Higher?

AXT has surged roughly 6,000% over the past year as investors target its indium phosphide wafer substrates, which are viewed as important for AI workloads and currently dominated by only a few suppliers. However, fundamentals lag the stock move: fiscal 2025 revenue fell to $88.3 million from $99.4 million in fiscal 2024, and the company posted $23.2 million in net losses while raising $550 million through dilution. Preliminary Q1 2026 revenue of $26 million-$28 million suggests some recovery, but the valuation remains stretched at 38x sales.

Analysis

AXTI is a classic scarcity trade masquerading as a fundamentals story: the market is discounting a multi-year capacity bottleneck in a niche input rather than current earnings power. That can persist longer than valuation discipline usually allows, but the setup is vulnerable because the re-rating has run far ahead of verified end-demand, leaving the stock exposed to any delay in Tier-1 qualification cycles or customer concentration disappointment. The second-order winner is not necessarily AXTI itself but the broader AI optical/interconnect stack, where faster data movement and lower power draw matter more than raw compute scaling.

The important read-through is that indium phosphide adoption is likely to be incremental, not substitutional. That favors a “pick-and-shovel” supply chain with high pricing power for the few qualified substrate vendors, but it also caps the TAM if hyperscalers continue to optimize around silicon and packaging improvements instead of wholesale material changes. In other words, the market may be overestimating the speed of adoption while underestimating how much incremental spend can flow to adjacent names with cleaner execution and less binary customer risk.

Near-term, the stock can stay elevated if management keeps signaling broader qualification wins, but the real catalyst window is 2-4 quarters, not days. The main downside trigger is evidence that the recent revenue inflection is inventory normalization rather than durable demand, especially if gross margin expansion fails to appear alongside top-line improvement. If that happens, the multiple compression could be sharp because the name is already priced for scarcity and growth simultaneously.

Contrarianly, the consensus may be missing that this is more a timing problem than a technology problem: the substrate thesis could be right while the equity is still too expensive to own outright. That argues for expressing the view through optionality or relative value rather than a naked long. The cleanest edge is to own beneficiaries of AI infrastructure scaling with lower narrative risk while fading the most crowded scarcity beta.

Key Decisions for Investors

• Avoid initiating a fresh outright long in AXTI at current levels; use only as a tactical trading vehicle until revenue and gross margin confirm a durable inflection over the next 1-2 quarters.
• Pair trade: long TSM vs short AXTI into any further AI-infrastructure optimism; TSM has clearer monetization of AI demand with materially lower execution and customer-concentration risk.
• For more aggressive expression, buy AXTI call spreads only on post-earnings pullbacks, targeting a 2-4 month horizon; the trade only works if Tier-1 qualification news arrives faster than the market expects.
• Relative-value basket: long NVDA/TSM, short AXTI as a hedge against AI hardware enthusiasm rotating from speculative materials to proven beneficiaries; this reduces binary risk while keeping AI exposure.
• If AXTI breaks lower on any commentary that suggests Q1 was inventory-led, cover quickly; the stock can de-rate 30-40% rapidly because the valuation is still far more sensitive to multiple compression than to near-term EPS.