The Real AI Trade: Data Center Semiconductors, Not Just GPUs
AI chip earnings strength in data center semiconductors refers to the recurring pattern where companies that build high-speed optical, RF and analog components for AI infrastructure demand are posting results above analyst forecasts, revealing that the supply chain behind artificial intelligence data centers is accelerating even when headline GPU makers face volatility. The key takeaway is blunt: investors watching only the obvious AI names are underestimating how durable and widespread the spending on AI infrastructure has become. Recent earnings seasons show that the biggest upside surprises are coming from the chips that move data, not the chips that do the math. That shift in leadership matters, because it hints that the AI buildout is broadening into a multi-year, system-wide cycle rather than a narrow bet on a handful of marquee processors.
MACOM’s Guidance Shock: Interconnect Chips Are Setting the Pace
MACOM’s latest quarter makes it hard to argue that AI infrastructure demand is cooling. The company beat fiscal third-quarter estimates and then stunned Wall Street with guidance far above what analysts had modeled, sending its shares higher on the news. This is not a story about GPUs. MACOM builds RF, optical and analog chips that carry data between processors inside data centers. Its next-generation modulator drivers sit in the high-speed interconnects hyperscalers need to link AI racks together for training and inference workloads. When cloud platforms add another rack of AI servers, someone must move the data between them at speed, and that is where MACOM’s products live. The striking part is that this is the second straight quarter the company has guided well above expectations, a pattern that signals a supply chain still tightening instead of cooling. In other words, the wiring behind AI is now a leading indicator for the entire cycle.
Applied Optoelectronics: Photonics Earnings Beats Ripple Through the Sector
Applied Optoelectronics’ second-quarter numbers reinforce the same message from another corner of the data center semiconductors universe. Sales rose 86% year over year and the company delivered an adjusted profit, beating analysts’ expectations. That upside triggered more than a double-digit premarket jump in its shares as the market rewarded its role as a photonics and data center components supplier. Importantly, the beat did not stay isolated: peers POET Technologies, Coherent and Lumentum all traded higher in sympathy, showing that investors increasingly treat photonics as a group trade. Applied Optoelectronics supplies high-speed optical transceivers, lasers and networking components that move data between AI servers. As hyperscalers ramp up AI infrastructure spending, investor interest in photonics companies has surged, fueled by expectations of a multi-year boom in optical connectivity that is seen as critical for scaling next-generation AI data centers. The market is signaling that the "optical plumbing" of AI is now a core earnings theme, not an obscure niche.
Tower Semiconductor: Silicon Photonics Sleeper Stock Signals Long-Term Investment
Tower Semiconductor shows how far the AI story now extends beyond the usual GPU suspects. It is not the obvious AI stock, and that is exactly why investors are looking at it now. The company sits in the quieter layer of AI infrastructure, making specialty chips for silicon photonics and silicon germanium platforms that help optical transceivers move information through AI data centers at high speed. Its exposure sits closer to the optical connection problem inside AI data centers than to the headline battles over processors. Recent expansion plans in Japan — focused on 300mm silicon photonics, silicon germanium and advanced packaging — amount to a concrete factory strategy that aligns with long-term demand for these chips. If you own AI infrastructure names, you should care about the wiring behind the rack as much as the processor inside it. The fact that Tower is being tagged as an AI data center sleeper stock is not hype; it is a recognition that sustained AI spending now includes the foundries building future photonics capacity.
Conclusion: AI Infrastructure Demand Is Broad, Deep and Still Accelerating
Taken together, MACOM’s aggressive guidance beats, Applied Optoelectronics’ photonics-driven upside and Tower Semiconductor’s expansion into silicon photonics chips all point in the same direction: AI infrastructure demand is broadening into the whole data center stack. The second straight quarter of MACOM guiding well ahead of Wall Street expectations signals a supply chain that is tightening rather than cooling. Investor interest in photonics companies has surged over the past year as hyperscalers step up AI infrastructure spending, with optical connectivity now seen as a critical enabler for next-generation AI data centers. The recent AI selloff in headline chip names only made Tower more interesting to investors, who are now willing to look beyond GPUs to the underlying optical connection problem. The market’s verdict is clear: if you focus only on processor vendors, you are missing where the most convincing AI chip earnings momentum is emerging — in the companies that keep the data flowing.
