Three 'Photon Tickets' are surging together. What on earth is this sector running on?

Recently, a reader asked me: POET, AEHR, and AXTI have all taken off. Is it a coincidence? Or is there a story we haven't understood?

Actually, it's not a coincidence. These three stocks represent three segments of the same industry chain—all are benefiting from the 'optical connectivity' cake of AI data centers. Let me put it in simpler terms:

AI computing power is no longer a battle of single chips, but a battle of 'interconnecting tens of thousands of GPUs.' Transmitting data between GPUs using electrical signals is no longer sufficient (too slow, too power-hungry, too hot), so a switch to optical signals is necessary. This is why giants like NVIDIA, AWS, and Microsoft are investing heavily in building 'Silicon Photonics' infrastructure.

But optical connectivity can't be handled by one company alone; it requires a three-tier division of labor:

First tier, substrate materials. For light to travel fast, the underlying material must be 'optically friendly'—Indium Phosphide (InP) is the key player. This is what AXTI does. AXT just announced a $550M public offering priced at $64.25 per share for 8,560,311 shares on April 21. All the funds will be used to expand the InP production capacity of its subsidiary (Beijing Tongmei Crystal). By the second half of 2026, capacity is expected to more than double. Northland raised its price target directly from $45 to $90, citing 'explosive demand for AI optical connectivity.' Q1 revenue guidance is $26–28M, and the financial report is due on April 30.

Second tier, optical chip packaging. With the substrate, electrical chips and optical chips need to be integrated together—this process is called 'photonic-electronic integration.' POET's Optical Interposer does this. POET was exposed on April 20 for accelerating its AI optical transformation, with its stock surging 10% in a single day and over 60% for April. The company still has $430M in cash, with a shipment target of over 30,000 units in 2026 (mass production in Q2). The board also approved the intention to list in the US (shareholders' meeting on June 26). Although it still reported a net loss of $42.7M on revenue of only $341K in Q4 2025, its commercialization inflection point is imminent.

Third tier, testing and burn-in. Chips need to be tested for reliability after production, especially for new technologies like silicon photonics, where customers must perform burn-in tests before mass production. AEHR is the leader in this area. It released its Q3 FY2026 earnings report on April 7—orders of $37.2M, a book-to-bill ratio of 3.5x (meaning new orders are coming in 3.5 times faster than shipments), and a backlog of $38.7M. More importantly, it secured a major silicon photonics testing order from a 'global network giant,' to be delivered in FY Q4 (before May 29). The stock price was $90.15 on April 20, exactly the shape of expectations being realized.

Looking at these three stocks together, the story becomes clear—substrate ($AXT(AXTI.US) ) → packaging ($POET Tech(POET.US) ) → testing ($Aehr Test(AEHR.US) ). AI optical interconnect is a complete industry chain; if any link gets stuck, no goods can be shipped.

I think the most interesting one here is AEHR. POET's commercialization realization is still on the way, AXTI just completed a large offering (dilution pressure suppresses the stock price short-term), only AEHR is a story of certainty where 'orders are already piled up and will turn into revenue in Q4.' AXTI has greater elasticity but also greater dilution risk; POET is a 'story stock within story stocks,' with amazing elasticity but one must beware of a pullback after good news is realized.

⚠️ These three stocks are not large in market cap (small caps), and their volatility will be much more intense than NVIDIA's. My view is: the sector is right, but which specific one to pick depends on your taste for 'certainty vs. elasticity.' I lean towards AEHR myself but understand friends who are bullish on POET's elasticity.