This Nvidia-Backed Company Just Entered the $154 Billion AI Photonics Market. Does This Make the $13 Stock a No-Brainer Buy?

Goldman Sachs projects AI-linked optical networking TAM to expand from about $15 billion in 2026 to $154 billion by 2028, with scale-up networking reaching $106 billion and co-packaged optics a $91 billion opportunity. Nokia is positioning for this shift, expanding photonic chip testing and packaging in Pennsylvania, while its Q1 2026 AI & Cloud segment grew 49% year over year, Optical Networks rose 20%, and it booked 1 billion euros in new AI infrastructure orders. The article is constructive on Nokia’s AI networking pivot, reinforced by recent insider buying and Nvidia’s $1 billion strategic investment in related collaborations.

Analysis

The setup is less about Nokia as a “telecom recovery” story and more about a structural bottleneck migration from compute to interconnect. That matters because the first-order winners in AI were obvious GPU vendors; the second-order winners are the picks-and-shovels around bandwidth density, packaging, and test capacity, where revenue can re-rate faster than end-market unit growth because the value content per rack is rising. If this thesis is right, the market will eventually stop paying only for AI capex growth and start paying for scarce, qualification-heavy infrastructure nodes that sit closest to the GPU supply chain.

The most important nuance is that photonics is still earlier in commercialization than the market tends to assume. Near-term demand can look explosive while actual monetization lags due to qualification cycles, interoperability issues, and customer concentration at hyperscalers; that creates a classic “orders now, earnings later” gap over the next 2-4 quarters. The risk is that consensus extrapolates TAM too aggressively before yields, reliability, and packaging economics are proven at scale, which could create air pockets in the names tied to AI networking if deployment schedules slip.

From a competitive standpoint, this is a positive read-through for any supplier with test/packaging depth and a negative one for companies dependent on legacy copper-based intra-rack connectivity. The more interesting second-order trade is that AI networking may become a margin capture battleground: if co-packaged optics wins, the value shifts toward fewer, more strategic component vendors, potentially compressing returns for commodity module assemblers. That said, the presence of strategic ecosystem investment from a large GPU incumbent increases the odds that this is an ecosystem buildout rather than a one-off customer win, which argues for staying long the enabling layer but selective on valuation.

Key Decisions for Investors

• Maintain / add to NVDA on weakness as a proxy for the ecosystem-enablement trade; use a 3-6 month horizon and prefer call spreads over outright stock to avoid paying full multiple for a crowded AI leader.
• Long GS vs. a basket of legacy telecom/equipment names over the next 6-12 months: GS benefits from monetizing the bottleneck narrative into capital markets flow, while the laggards remain exposed to slower optical conversion cycles.
• Initiate a small tactical long NOK only on post-rally consolidation; size it as a speculative 2-4 quarter trade with a tight stop, because the upside is tied to qualification wins while downside is multiple compression if AI orders decelerate.
• Pair trade: long optical/photonics enablers and short a basket of copper/interconnect or generic network hardware exposure for 6-9 months; the trade works if interconnect spend shifts from bandwidth volume to bandwidth density.
• If NOK continues to gap on headlines, use call spreads or collars rather than cash equity; the risk/reward is now asymmetric to the downside after the run-up, while implied volatility should remain elevated on every new partnership announcement.