Optics and photonics should be among AI's biggest winners as faster GPU networks need optical connections that suppliers cannot expand quickly enough. Lumentum says some components are sold out through early 2029; Sivers is investing $30M to target more than 100M lasers a year, with operations expected in Q4 2027. Potential US restrictions on Chinese transceivers could reshape supply in 2027. Applied Optoelectronics and Sivers are the favorites. x.com  x.com 

Applied Optoelectronics, Lumen and Coherent are the obvious beneficiaries of a US draft ban on Chinese data center components, reported by Reuters, and are already gapping up. x.com 

Enterprise data centers and AI racks are competing for limited parts, benefiting data-center OEMs, ODMs, CPU, memory and storage suppliers. Even a modest AI slowdown would not clear the shortages; customers will order ahead to secure allocation. Dell, HPE, Lenovo, Foxconn, Micron, SK hynix, Samsung, SanDisk, AMD and Intel are named beneficiaries. x.com 

AI infrastructure bottlenecks could give eight stocks as much as 18x upside. Nvidia supplies frontier-model GPUs; Broadcom and Marvell supply custom silicon and networking; TSMC makes advanced chips; Micron supplies sold-out HBM; Vertiv provides power and cooling; Palantir monetizes enterprise AI; Alphabet spans TPUs, research, cloud and apps. The list tracks where AI infrastructure spending is flowing. x.com  x.com 

Western lidar makers Ouster and Aeva stand to fill the vacuum if Chinese rival Hesai is banned in the West. Hesai faces US national-security scrutiny over its expanded Nvidia partnership, after warnings that its sensors could be disabled or exploited remotely; a Hesai firmware update disabled lidar units on February 29. Upstream laser suppliers to Western lidar, such as Lumentum and Sivers, would also gain revenue. The news may not matter much while markets are crashing. x.com  x.com 

Lumentum and Coherent sit at the center of a multiplicative optical bottleneck, where parts, capacity, yield and customer sign-off each limit output. Next-gen AI clusters need optical interconnect, and McKinsey sees 800G transceiver output 40-60% short of demand through 2027 as the market doubles to $26B in 2026. Nvidia invested $2B each in both to lock in laser supply, and the FCC is drafting curbs on Chinese transceiver imports; Chinese suppliers hold ~60% of optical-component revenue. x.com  … Show more

Networking is the next AI hardware trade after memory, whose share of rack value rose from 15% to 35%. Optics is moving inside the server, as Nvidia's Feynman architecture builds in co-packaged optics and silicon photonics. Scale-across networking could grow 30% a year just linking existing data centers, as inference moves closer to users and training spreads across sites. Supply is tight: Lumentum's CEO said it can meet only 30% of expected 2027 demand and sees optics tighter than memory. x.com 

These ten companies span every layer of AI capex, which Goldman sees surging to $1.2 trillion in 2027, so each should make money. Credo, Broadcom, Micron, Nvidia, TSMC, AMD and Amkor cover connectivity, custom chips, memory, GPUs, foundry and packaging; Broadcom expects AI revenue of about $115B in fiscal 2027. Vistra supplies power under 20-year nuclear contracts. IREN and Nebius rent AI compute, with Nebius targeting $7-9B run-rate revenue by year end. x.com  x.com  x.com  x.com  x.com  x.com  x.com  … Show more

Critical-materials suppliers offer exposure to a major bottleneck this decade: AI, data centers, defense, robotics and electrification should drive massive demand across copper, graphite, rare earths, nuclear fuel and other inputs.
hypertechinvest on X

SpaceX's $40B financing for GPU procurement adds another large, funded source of AI-infrastructure demand, with spending flowing into GPUs, memory, networking and optics. At the same time, InP substrates are seriously short, and Morgan Stanley identifies InP, EML and CW lasers as primary bottlenecks; suppliers that secure raw materials should gain pricing power. Any slowdown looks unlikely. x.com