MarketMoodz
Nvidia Bets Billions on Silicon Photonics to Power AI
Nvidia has committed at least $6.5 billion to silicon-photonics investments since March, deploying roughly $2 billion into Lumentum, Coherent and Marvell and pledging $500 million to Corning, CNBC reports. The push aims to shift data movement inside and between data centers from copper to light, increasing bandwidth and lowering energy use to scale larger AI models and GPU clusters.
Key Takeaways
- CNBC reports Nvidia committed at least $6.5 billion to photonics investments since March, with about $2 billion already deployed to Lumentum, Coherent and Marvell.
- Nvidia pledged $500 million to Corning to develop advanced optical connectivity and participated in Ayer Labs' $500 million Series E round alongside AMD.
- CEO Jensen Huang said at GTC that silicon photonics capacity needs are far higher than current global supply, underscoring long-cycle demand.
- Photonics can cut energy use and scale bandwidth by moving data with light instead of copper, but manufacturing yield and alignment challenges may slow broad adoption until ~2028.
- Photonics beneficiaries have seen large stock moves in 2026 — Lumentum +134%, Coherent +96%, Marvell +122%, Corning +111% — according to CNBC’s figures.
People Involved
- Jensen HuangNvidia CEO
Entities Involved
- Nvidia (NVDA)Lead investor and driver of photonics strategy for AI infrastructure
- Lumentum (LITE)Photonics component supplier; recipient of Nvidia investment
- Coherent (COHR)Photonics company; recipient of Nvidia investment
- Marvell (MRVL)Semiconductor and connectivity supplier; recipient of Nvidia investment
- Corning (GLW)Optical-fiber and connectivity manufacturer; recipient of a reported $500M pledge
- Ayer LabsPhotonics startup that raised a reported $500M Series E including Nvidia participation
- Advanced Micro Devices (AMD)Participant in Ayer Labs round alongside Nvidia
- nEyePhotonics company backed in an $80M Series C by Alphabet and Microsoft venture arms
- Alphabet (GOOGL)Venture investor in photonics ecosystem (nEye Series C)
- Microsoft (MSFT)Venture investor in photonics ecosystem (nEye Series C)
- TeramountPhotonics-related company linked to Nvidia ecosystem investments
- Celestial AIAI-infrastructure company tied to Nvidia’s ecosystem moves
- EnosemiAlleged AMD acquisition in 2025 (unverified)
MarketMoodz Analysis
For investors, Nvidia’s multi-billion-dollar run on silicon photonics signals a potential structural shift in AI infrastructure demand. If the reported $6.5 billion cadence holds, suppliers of lasers, modulators, photonic integrated circuits and specialty fiber could see multi-year order streams tied to GPU-cluster buildouts and cloud data-center upgrades. That would push revenue visibility for listed photonics suppliers and nearby component makers, while also compressing total cost of ownership for hyperscalers through lower energy bills and higher rack-level bandwidth.
The technology path is familiar to telecom’s shift from electrical to optical links, but silicon photonics adds manufacturing complexity: tight alignment, wafer-scale yields and packaging remain obstacles. Industry estimates and Nvidia executives (per CNBC) flag large-scale adoption risks extending to about 2028, which implies a multi-year transition with episodic supplier wins and capacity bottlenecks. Stocks of reported recipients — Lumentum, Coherent, Marvell and Corning — have already rallied sharply in 2026, reflecting market expectations for outsized demand; investors should treat those moves as anticipatory and watch guidance in quarterly reports for sustained order flow.
Next up: verify cadence and timing. Watch Nvidia’s capital-allocation disclosures and partner filings for confirmed commitment amounts, supplier order books and milestone-based payments. Track yield improvements and pilot deployments from Ayer Labs, Teramount and other partners, plus cloud-provider procurement cycles — those metrics will determine whether photonics becomes a near-term revenue driver or a longer, strategic play that only meaningfully reshapes AI compute economics after 2027.
Source: Original Article