Analyst(s): Brendan Burke
Publication Date: March 3, 2026
NVIDIA has announced parallel strategic partnerships with Coherent Corp. and Lumentum Holdings, investing $2 billion in each company alongside multibillion-dollar purchase commitments to advance silicon photonics and optical interconnect manufacturing for AI data centers. The dual-vendor structure reflects a deliberate supply chain diversification strategy that positions optics as the emerging gating factor on continued AI infrastructure scaling.
What Is Covered in This Article:
- NVIDIA’s $4 billion combined investment in Coherent and Lumentum for optical R&D and manufacturing
- Indium phosphide and laser fabrication as the emerging constraint on AI infrastructure scaling
- Parallels to NVIDIA’s prior supply chain investments in HBM, CoWoS, and networking
- The distinct roles of scale-up and scale-out optics in next-generation AI clusters
- Implications for the competitive optical ecosystem and the copper-to-photonics transition
The News: NVIDIA announced on March 2, 2026, two separate multiyear strategic partnerships with Coherent Corp. and Lumentum Holdings to advance optical interconnect technology for next-generation AI data centers. Each agreement includes an NVIDIA investment of $2 billion, alongside a multibillion-dollar purchase commitment and future access to capacity, supporting Coherent and Lumentum in expanding U.S.-based manufacturing and research and development for advanced laser and optical networking products.
Coherent CEO Jim Anderson described the partnership as an expansion of a 20-year relationship with NVIDIA across multiple product families, while Lumentum CEO Michael Hurlston noted the company is investing in a new fabrication facility to increase capacity.
Jensen Huang, founder and CEO of NVIDIA, stated in the Coherent announcement: “With Coherent, NVIDIA is pioneering next-generation silicon photonics to enable AI infrastructure at unprecedented scale, speed, and energy efficiency.”
In the Lumentum announcement, Huang added: “Together with Lumentum, NVIDIA is advancing the world’s most sophisticated silicon photonics to build the next generation of gigawatt-scale AI factories.”
NVIDIA’s $4B Optics Bet Signals Photonics as AI’s Next Bottleneck
Analyst Take: NVIDIA’s decision to simultaneously invest $2 billion in each of two optical component suppliers represents a signal that NVIDIA views optics as a strategic bottleneck requiring the same level of supply chain engineering it has applied to high-bandwidth memory (HBM), advanced packaging, and networking over the past several years. NVIDIA’s supply chain decisions effectively set the capacity ceiling for the entire AI infrastructure buildout, and that ceiling is now being defined by photonics.
Indium Phosphide Is the New CoWoS
The core constraint NVIDIA is addressing is the limited global capacity for indium phosphide (InP) laser fabrication — the compound semiconductor material required for the high-speed continuous-wave lasers used in 800G and 1.6T optical transceivers and co-packaged optics (CPO) engines. Unlike silicon wafer production, which benefits from decades of scale-driven cost reduction and a broad foundry ecosystem, InP epitaxial growth is performed by a small number of specialty manufacturers with inherently lower throughput and yield. Current transceiver demand is reported to exceed supply by a factor of two due to InP constraints, creating a dynamic that mirrors the CoWoS advanced packaging bottleneck of 2023–2024 that constrained NVIDIA’s H100 production ramp.
Futurum’s base-case forecast assumes moderate bottlenecks in HBM and advanced packaging through 2027, before supply/demand balance returns by 2028–2029. Optical supply chain constraints could follow a similar multi-year resolution timeline. By investing directly in Coherent’s and Lumentum’s manufacturing capacity and R&D, NVIDIA is attempting to compress that timeline and applying the same playbook it used when it pre-paid for CoWoS capacity, cultivating multi-source HBM relationships with SK Hynix, Samsung, and Micron to pull supply forward against surging demand.
Dual-Source Strategy Reflects Lessons From Prior Bottlenecks
The decision to invest in two optical partners rather than selecting a single source is architecturally deliberate rather than financially redundant. Coherent and Lumentum occupy complementary positions in the optical value chain. Coherent brings established expertise in CPO integration, fiber-to-chip connectors, and broader optical packaging. Lumentum specializes in high-power continuous-wave laser chips used in external laser sources for CPO.
This complementarity means NVIDIA is building depth across distinct layers of the optical stack, from the InP laser source through to the packaged optical engine. The $4 billion in combined investment also provides Coherent and Lumentum with the balance-sheet certainty to commit to new fabrication lines and U.S.-based manufacturing expansion.
Scale-Up and Scale-Out Optics Are Distinct Engineering Challenges
NVIDIA’s press language references both “optical interconnects and advanced package integration” and “large-scale AI networks,” which map to two distinct optical domains with different technical requirements. Scale-up optics — the chip-to-chip interconnects within NVIDIA’s NVLink domain — are approaching a transition point as the NVLink-connected GPU domain expands from 72 GPUs in Blackwell to larger configurations in Rubin, pushing copper electrical signaling toward its physical limits in bandwidth-per-watt at the frequencies required for 800 Gbps links. NVIDIA’s CPO-based systems, such as Quantum-X and Spectrum-X Photonics, target power consumption reductions of up to 3.5x and resiliency improvements of 10x, gains that become essential when a single AI rack draws up to 600 kW and optical networking alone can consume 10% of that power envelope. Scale-out optics — the rack-to-rack fabric layer using pluggable and eventually co-packaged transceivers — face a volume problem, as clusters scaling to tens of thousands of GPUs require hundreds of thousands of optical links, each dependent on the same constrained InP laser supply. NVIDIA is positioning itself to lead both domains by transitioning from traditional pluggable transceivers to vertically integrated CPO.
The Copper Cliff Accelerates the Optical Imperative
The timing of these investments is driven by the physical limits of copper interconnects at the data rates AI clusters now demand. Traditional copper links are impractical for the 800 Gbps speeds required in modern AI fabrics, where switches are positioned at the end of a row to create low-latency topologies, and nearly every server-to-switch link now requires an optical connection. As copper reaches its limit at higher frequencies, the optical transition is not optional but mandatory, and the investment in Coherent and Lumentum ensures NVIDIA controls access to the specialized laser and packaging technology required to execute that transition.
Major hyperscaler customers such as Meta and Microsoft have separately advocated for a competitive ecosystem of pluggable CPO to avoid vendor lock-in, and the nonexclusive structure of both agreements, combined with Coherent’s work on high-performance sockets that bridge pluggable modules and CPO, suggests NVIDIA is positioning its optical supply chain to support both proprietary and interoperable configurations, preserving flexibility as industry standardization efforts mature.
What to Watch:
- Coherent’s and Lumentum’s ability to ramp InP laser fabrication within 18–24 months will determine whether optical supply keeps pace with NVIDIA’s Rubin-generation GPU production timeline.
- Hyperscale customers advocating for pluggable CPO interoperability could force architectural compromises that reshape how open or proprietary NVIDIA’s optical interconnect layer becomes.
- Broadcom’s Tomahawk 6 CPO and Arista’s linear pluggable optics represent competing architectural approaches that may either validate NVIDIA’s silicon photonics direction or fragment industry standardization.
- China’s concentration of global indium supply introduces a geopolitical risk that U.S.-based manufacturing investment alone cannot fully resolve without diversified upstream materials sourcing.
See the complete press releases on NVIDIA’s strategic partnerships with Coherent and Lumentum on the NVIDIA website.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other Insights from Futurum:
Coherent Q2 FY 2026: AI Datacenter Demand Lifts Revenue and Margins
Did SPIE Photonics West 2026 Set the Stage for Scale-up Optics?
Will NVIDIA’s Meta Deal Ignite a CPU Supercycle?
Image Credit: Coherent
Author Information
Brendan is Research Director, Semiconductors, Supply Chain, and Emerging Tech. He advises clients on strategic initiatives and leads the Futurum Semiconductors Practice. He is an experienced tech industry analyst who has guided tech leaders in identifying market opportunities spanning edge processors, generative AI applications, and hyperscale data centers.
Before joining Futurum, Brendan consulted with global AI leaders and served as a Senior Analyst in Emerging Technology Research at PitchBook. At PitchBook, he developed market intelligence tools for AI, highlighted by one of the industry’s most comprehensive AI semiconductor market landscapes encompassing both public and private companies. He has advised Fortune 100 tech giants, growth-stage innovators, global investors, and leading market research firms. Before PitchBook, he led research teams in tech investment banking and market research.
Brendan is based in Seattle, Washington. He has a Bachelor of Arts Degree from Amherst College.
