cryptooo

🚨Why I think $MRVL acquisition of Celestial AI is a MASSIVE tailwind for $POET technologies Marvell just shook the semiconductor stack. The Celestial AI acquisition, $3.25B upfront, potentially $5.5B, is more than a headline. It’s a direct swing at the memory wall. A very loud message that the next phase of AI scaling runs on light, not electrons. Credits to Marvell for locking down the Photonic Fabric. Fair. They NEEDED that IMO. It’s one of the most important architectures in the industry. But if your analysis stops there, you’re overlooking the second-order name with far more torque: POET Technologies $POET Marvell paid to solve a physics problem... Celestial’s architecture is elite. Silicon Photonics, however, comes with a fundamental limitation. It cannot generate its own light. GPUs run at temperatures that turn integrated lasers into toast. So the system needs a remote light engine pushing photons in through fiber. Marvell now owns the chassis but they still require the engine This Is Where POET Fits in... This is supply-chain reality Celestial has been working with POET since 2022 for that exact component. The product - POET’s Starlight. The advantage - up to 75% lower cost based on Celestial’s own presentation. The status - designed in and validated. Marvell acquired Celestial’s architecture along with Celestial’s dependencies. One of my first thoughts when I saw this was, will Marvell replace Poet and use their current laser tech (Inphi) instead? Possible in theory. In practice, I dont believe there is a runway for a replacement path. Marvell guided investors to revenue in 2H 2028. In hardware timing, that is already around the corner. The architecture is fixed. Production planning is underway. Re-qualifying a new light-engine vendor now risks delaying the entire roadmap that justified the $3.25B check. They require a wafer-scale, passive-alignment solution that exists today. POET holds that position. So what is this hypothetically worth? Let’s be extremely grounded and look at Marvell’s own guidance for this deal. Marvell is targeting a $500M revenue run-rate by late 2028, doubling to $1B by 2029. If we apply standard industry napkin math: The light engine (laser source) typically makes up roughly 10-15% of the hardware BOM in optical systems. If POET holds the socket (and I believe they do) the math gets interesting: At a conservative 10%, for $500M platform revenue, we get a ~$50M potential revenue to the light supplier. At $1B platform revenue, ~$100M potential revenue. For a company like POET currently doing <$5M a year, a single contract ramping to $50M+ ARR is a fundamental re-rating event. Note: This is strictly a hypothetical model based on BOM assumptions, not official guidance. But the volume implications are undeniable. Separately... With Celestial off the table, the list of independent companies with proven optical-interposer IP is shrinking. Broadcom, AMD, Nvidia are all watching. The buy-vs-build equation shifts toward whoever is left standing. POET transitions from interesting small-cap to strategic scarcity play. TLDR; Marvell’s move confirms the broader trend. Copper is reaching its limits inside the rack. Light becomes the growth curve. And by validating optical scale-up, Marvell effectively elevated POET into a Tier-1 supplier lane for the AI era. The optical supercycle is already underway. This is not a recommendation for anyone. Nor is it financial advice. There is still ALOT of uncertainty with POET and alot that needs to be proven, but the story for me keeps getting better and better.