POET Technologies: The Missing Piece In Marvell's AI Empire

This is less a near-term earnings story than a credibility inflection: once a component supplier moves from design-win narrative to shipping 1.6T systems, the market tends to re-rate on revenue visibility rather than unit economics. The second-order winner is likely the broader co-packaged optics / photonics supply chain, because successful qualification at one hyperscaler tends to create a template for rapid replication across adjacent customers. The key loser is any slower-moving interconnect incumbent whose roadmap depends on incremental copper optimization rather than a step-function bandwidth jump.

The critical catalyst is not initial shipment volume, but the speed of qualification expansion over the next 2-4 quarters. If these systems get pulled into a mainstream AI platform cycle, revenue can inflect well before headline demand catches up, because hyperscalers typically pre-buy infrastructure to secure supply ahead of node transitions. That said, the entire bull case is brittle to execution slippage: a one- or two-quarter delay in yields, thermal performance, or integration milestones would push out the re-rating window and compress multiple expansion.

The market is probably underestimating how much of the upside is optionality versus base case. A sub-$1B company moving toward a multi-bagger outcome can look cheap on 2027 revenue forecasts, but the path matters more than the destination: if POET becomes a strategic enabler inside a larger ecosystem, takeout value rises materially before fundamentals fully show up. For NVDA, the impact is indirect but real: faster interconnect adoption extends the usable economics of next-gen GPU deployments, supporting capex intensity and reducing bottlenecks that could otherwise limit Rubin-scale rollouts.