Key Takeaways
- • Nvidia commits $6.5 billion to photonics investments since March 2026, including $2 billion each to Coherent and Lumentum, $2 billion to Marvell, and up to $3.2 billion to Corning, plus $500 million participation in Ayar Labs' Series E funding round.
- • $15 billion invested in photonics sector since January 2026 signals industry-wide recognition that copper interconnects have reached physical bandwidth and power consumption limits for AI data center clusters.
- • Copper wiring has hit fundamental physics constraints as AI training clusters demand exponentially higher bandwidth; silicon photonics using light-based transmission offers substantially higher bandwidth at lower power consumption than electrical signals.
- • Photonics supply chain faces critical capacity shortage through 2028, with foundries locking down silicon photonics production capacity as AI infrastructure scaling accelerates, creating a structural bottleneck that Nvidia's strategic investments aim to resolve.
- • Multiple photonics companies receiving major capital commitments include established optical component makers Coherent, Lumentum, and Marvell (post-Celestial AI acquisition), indicating this transition is not a single-company phenomenon but an industry-wide infrastructure pivot.
Summary
Nvidia has emerged as the dominant investor in the silicon photonics sector, committing at least $6.5 billion since March 2026 to replace copper wiring in AI data centers. The investment strategy reflects a critical acknowledgment across the industry that copper interconnects have reached their physical limits precisely when AI training clusters are demanding exponentially higher bandwidth. Photonics technology, which uses light rather than electrical signals to transmit data, offers substantially higher bandwidth capacity at significantly lower power consumption—two constraints that become mission-critical when thousands of GPUs must operate as a unified system.
The capital deployment demonstrates strategic depth: Nvidia invested $2 billion each in optical component makers Coherent and Lumentum in early March 2026, with both deals including multi-billion-dollar purchase commitments and funding for new U.S. fabrication capacity. An additional $2 billion went to Marvell Technology, which acquired photonics startup Celestial AI in December 2025 and is actively developing silicon photonics solutions for AI networking. Corning received up to $3.2 billion in commitments, while Ayar Labs secured $500 million in Series E funding with Nvidia's participation. This $6.5 billion figure sits within Nvidia's broader $40 billion equity investment portfolio focused on AI infrastructure in 2026.
Industry-wide photonics investment has reached $15 billion since January 2026, signaling that the transition from copper to light-based interconnects is no longer speculative but rather an accepted necessity. The bottleneck driving this acceleration is not theoretical: the global AI infrastructure boom has collided with what industry participants call the "copper wall"—a fundamental physics barrier where conventional electrical interconnects cannot deliver the bandwidth density required for next-generation AI clusters. Power consumption from moving electrons through copper wiring increasingly dominates data center operating costs and limits cluster scalability.
A critical constraint emerging in this transition is supply-side capacity. Foundries have locked down silicon photonics production capacity through 2028, reflecting aggressive demand forecasting by AI infrastructure builders. This supply shortage creates both risk and opportunity for investors: companies positioned in the photonics supply chain face multi-year order backlogs, while AI infrastructure operators must secure long-term supply agreements to meet scaling timelines. Nvidia's strategy of direct investment in component makers and foundry capacity appears designed to secure supply certainty while enabling manufacturing scale-up necessary to support the industry's exponential growth trajectory in AI infrastructure deployment.
The capital deployment demonstrates strategic depth: Nvidia invested $2 billion each in optical component makers Coherent and Lumentum in early March 2026, with both deals including multi-billion-dollar purchase commitments and funding for new U.S. fabrication capacity. An additional $2 billion went to Marvell Technology, which acquired photonics startup Celestial AI in December 2025 and is actively developing silicon photonics solutions for AI networking. Corning received up to $3.2 billion in commitments, while Ayar Labs secured $500 million in Series E funding with Nvidia's participation. This $6.5 billion figure sits within Nvidia's broader $40 billion equity investment portfolio focused on AI infrastructure in 2026.
Industry-wide photonics investment has reached $15 billion since January 2026, signaling that the transition from copper to light-based interconnects is no longer speculative but rather an accepted necessity. The bottleneck driving this acceleration is not theoretical: the global AI infrastructure boom has collided with what industry participants call the "copper wall"—a fundamental physics barrier where conventional electrical interconnects cannot deliver the bandwidth density required for next-generation AI clusters. Power consumption from moving electrons through copper wiring increasingly dominates data center operating costs and limits cluster scalability.
A critical constraint emerging in this transition is supply-side capacity. Foundries have locked down silicon photonics production capacity through 2028, reflecting aggressive demand forecasting by AI infrastructure builders. This supply shortage creates both risk and opportunity for investors: companies positioned in the photonics supply chain face multi-year order backlogs, while AI infrastructure operators must secure long-term supply agreements to meet scaling timelines. Nvidia's strategy of direct investment in component makers and foundry capacity appears designed to secure supply certainty while enabling manufacturing scale-up necessary to support the industry's exponential growth trajectory in AI infrastructure deployment.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.