X-FAB Silicon Foundries' exposure to co-packaged optics and 800V power is now confirmed, but it is a 2028 patience trade rather than a near-term catalyst. Per @aleabitoreddit's earnings-call recap, X-FAB said it is making progress in CPO and expects the 800V transition to support SiC and GaN demand. Photonics volume production and the new Erfurt fab both land at the end of 2028, and wide-bandgap revenue dipped 30% quarter on quarter on one customer's timing. x.com 

X-FAB's silicon carbide and photonics growth will become impossible to ignore once autos recover, especially at 1.29x book value. @aleabitoreddit notes that Nvidia's 800V DC push favors GaN and SiC power chips, that fabless beneficiaries such as Navitas rely on foundries, and that buyers increasingly want Western supply chains. x.com 

X-FAB still has a lot to prove because its execution has consistently been poor, despite promising technology. Management targeted about €300 million of silicon carbide revenue by 2026 and delivered essentially nothing, with most of that business going to Infineon. Its new photonics story, led by a €48 million EU consortium with NVIDIA as a partner, makes headlines, but NVIDIA has no obligation to become a customer. It stays on the watch list until actual execution under the new CEO. x.com 

X-FAB is a good name because 800V demand is booming for its traditional CMOS business, not because of CPO, which is too small and far from taking off. It can shift capacity from unprofitable automotive to profitable cloud-provider work, and its silicon carbide also targets 800V. It is a rare fab with products at about 1.5x book, cheap because past profits were weak; earnings should improve as Chinese foundry costs have doubled. A 2–2.5x book multiple looks achievable, but not a 5–10x return. … Show more

X-FAB is a pure-play silicon carbide and gallium nitride foundry, reminiscent of an early Tower Semiconductor or Soitec. Per @aleabitoreddit, NIST filings name it the only high-volume SiC foundry in America, making it critical infrastructure as NVIDIA pushes 800 VDC and power-chip peers Navitas and Wolfspeed rerate hard. It is also in CHIPS Act 2 blueprints for photonics, with Nokia and NVIDIA evaluations underway and volume production expected in H1 2028, all at a $1.5B market cap. x.com 

X-FAB's celebrated AI capacity expansion means very little for shareholders today: no new revenue, no new customer and no meaningful earnings before 2029. It is merely the formal handover of a €127 million subsidy approved in late 2025, a normal 30-40% rate, and X-FAB must still fund most of the investment. The cleanroom will not produce until late 2028 and mainly serves its legacy automotive, smartphone and medical markets, making it one of the least exciting AI-linked stocks fundamentally. … Show more

Fabrinet deserves an Outperform rating and an $850 December 2026 target as an irreplaceable precision maker of 800G and 1.6T optical interconnects, AI's binding constraint. HPC revenue jumped from $15M to $86M in a quarter, new capacity serves orders in hand, and Q3 FY26 guidance of $1.15-1.20B implies about 35% growth after seven straight EPS beats. NVIDIA and Cisco concentration is a risk, but switching costs run upstream. Legacy telecom and automotive remain a drag. x.com 

Credo's selloff on a raise reflects impatience, since its next growth leg is a fiscal 2028 story that begins in May 2027. It printed $479M, guided $525-535M and lifted full-year growth to more than 85%, yet sold off for a second straight print. CEO Bill Brennan called fiscal 2027 just a stepping stone for optics: OmniConnect, micro-LED cables and near-package optics turn on in fiscal 2028, when AECs step up to 1.6T. x.com  x.com 

X-FAB looks quite undervalued, with 2x potential in the short term, as a Bernstein upgrade and new CHIPS Act funding arrive. Bernstein upgraded it to Outperform and raised its target to €12.8 from €5, leaving about 38% upside from roughly €9.3. Per @aleabitoreddit, X-FAB receives €127.4 million under the CHIPS Act for its Fab4Micro MEMS foundry in Erfurt, with production planned for 2028. x.com  x.com 

Applied Optoelectronics' case now hinges on visible production progress. Its drop on OpenAI's revenue run-rate figures reflected an accounting question, not any change in laser or memory supply, per Carrio Research. If AAOI fails again next quarter and pushes its timelines out another quarter, it will be ugly. x.com