Today, Smart Eye unveils a world-first multimodal in-cabin AI agent. It goes beyond voice and vehicle data to understand the driver, their gaze and state, others inside, and what’s happening outside. See it at #InCabinEurope, booth #305. Read more: hubs.ly/Q04y5V4z0

$GFS & $MRVL EXPAND U.S. CAPACITY FOR AI OPTICAL NETWORKING GlobalFoundries and Marvell signed a multi-year agreement to expand silicon germanium manufacturing capacity at GF’s Burlington, Vermont facility. The added capacity will support Marvell’s next-gen pluggable optics, near-packaged optics and co-packaged optics as AI data centers push toward higher-bandwidth interconnects. GF’s current SiGe platform supports 200G-per-lane connectivity, with a roadmap toward faster generations. Financial terms were not disclosed.

Glad $SIVE is attending CIOE Shenzen and hope my followers can go too! If you want my read on current landscape: -> Chinese pluggable makers like CIG (剑桥科技) are facing severe shortages in 70mW-200mW lasers. If you look at filings from CIG, they claim "substantially longer delivery times and deposits to secure capacity” along with statements of shortages. -> Innolight and others were also looking to secure multiple laser suppliers with LTAs recently from their filings. On the US side of things: -> $COHR was a major merchant laser supplier before, but largely withdrew from the market after rerouting them internally. "I do not see any time in the near future where we would be selling Indium Phosphide lasers externally." From their ER transcript. And then likely turned into a buyer in an already supply constrained market... where "over the long term, we will have some portion of our datacom transceivers that will be supported by external sources." -> $AAOI not really known to be a merchant supplier, but said same thing about laser capacity being rerouted towards transceivers. And having to turn away customers for lasers. -> For $LITE, they still supply lasers but stated: "We are commanding a significant price premium" for CW lasers $LITE Wupen Yuen: "wherever they can get the laser source, they will use that solution to support their build-out." So hinting... about customers just finding anything available due to shortages. -> $MTSI has no meaningful capacity online now, but they're already stating customers are approaching them with urgency to secure CW laser supply. -> Trendforce reported $AMD and hyperscaler CSPs are aggressively going out to secure CW laser supply to avoid future bottlenecks. And then you have Europe... Where $SIVE is coming online with "tremendous capacity available now" from their foundry partners. With 100M+ laser capacity Q4 2027, and maybe if you look at their 2:1 ratio implying ~200M targeted laser capacity from external foundries. They also happen to offer the same power range (70mW, 100 mW, 200mW) as the ones currently in shortage by pluggable makers in China. There's a massive void to fill, so this dramatically increases the chance of converting customers. (esp. Supported by Lumentum statements) My speculation was that some of the 6 active pluggable engagements were from China? Which is why they're attending the conference. So to any of my Chinese followers, maybe you can ask a few questions like: - if Europe is a laser source geography, if Innolight/Eoptolink/Cambridge are considering $SIVE. - if they're seeing ASP hikes in lasers, and what CW products are hardest to obtain If you are attending CIOE in Shenzhen!

Holy fuck

@AurelionRsch sat down with $LITE's VP of Investor Relations. Their meeting notes came to me, and I added a device engineer's comments on top. The management access is theirs. The technical layer is mine. The result went up today, fully free. photoncap.net/p/a-conversati… IR answers are usually true. The question is how far each true statement actually reaches. Take the sentences from that call apart at the product level and this is what falls out: 1. Capacity is up eight fold in two and a half years, yet everything is still sold out. Wafer starts and top bin yield are different numbers 2. "Few can make high power lasers" is true, but the hard part is not the milliwatts. Four terms explain where it actually lives: RIN, COD, SMSR, FIT 3. If CPO slips, does Lumentum get hurt, or do pluggables just live longer? The answer depends on why it slips 4. Is Nvidia concentration a risk or a snapshot? Second sourcing is already on the public record