Nvidia Targets 20x InP Laser Capacity Through 2030
Nvidia has asked its key suppliers to lift indium phosphide laser capacity by as much as twentyfold by 2030 to support the networking needs of increasingly dense AI clusters. The objective is to move data faster across thousands of GPUs and keep pace with the rapid expansion of AI workloads in modern data centers. Early feedback from suppliers suggests a best-case uplift of around 12x, but a gap remains as demand accelerates and builders push for denser interconnects.
Optics Bottleneck Could Slow AI Rollout
Industry watchers say optics—specifically the hardware that carries data between accelerators—could become the next critical constraint as AI models scale. Co-packaged optics, which place networking components closer to the chips, offer speed and efficiency gains but rely on a steady, substantial supply of indium phosphide lasers and related photonics gear.
- InP laser capacity target: 20x by 2030 (Nvidia).
- Supplier response: roughly 12x uplift currently offered.
- InP optics market forecast: from about $1.9 billion in 2025 to roughly $22.75 billion by 2030.
- Capacity projections: Lumentum 9.0 billion dollars by 2030; Coherent about 4.3 billion; Broadcom around 4.5 billion.
- Analyst perspective: Rosenblatt Securities emphasizes Coherent and Lumentum as likely beneficiaries from efficiency gains and transceiver market expansion.
The push comes as Nvidia fears that networking constraints could bottleneck AI deployment as cluster sizes scale rapidly. Adopting co-packaged optics to bring connections closer to the processor is central to boosting throughput while reducing power draw, but it also heightens reliance on a narrow group of photonics suppliers.
Investor Take: What It Means For Stocks
For investors, the optics squeeze feeds a broader narrative about scarce hardware in AI ecosystems. The phrase stocks bottleneck soar 1,000% has entered market chatter as traders recalibrate expectations around when and where bottlenecks will ease. If optics makers can scale capacity quickly and cost effectively, share prices of photonics players could outperform as AI deployments accelerate across hyperscale data centers.
- Key beneficiaries could include LUMENTUM, COHERENT, and Broadcom, which sit at the intersection of high-speed optics and data-center interconnects.
- Any sustained upside will hinge on turning announced capex into timely, usable capacity without eroding margins.
- The 2030 horizon means investors should monitor capital expenditure cycles, yield improvements, and supplier diversification in photonics.
Market Context: A Hardware-Heavy AI Era
AI expansion remains unusually hardware-driven. Cloud providers and hyperscalers continue pouring capital into GPU fleets, high-bandwidth interconnects, cooling systems, and power infrastructure. In that sense, optics is no longer a back-office afterthought but a central piece of the AI hardware stack. With data center architectures evolving, the ability to move massive data streams quickly and efficiently is as important as the raw compute power itself.
As the optics market grows, the race to secure supply lines becomes a major determinant of project timelines and ROI. If Nvidia and its peers can secure a stable, scalable supply of InP lasers and related components, the path from lab to deployment accelerates—and markets reward suppliers that can deliver.
Data Snapshot: What to Watch This Quarter
- Nvidia demand: aim for 20x InP laser capacity growth through 2030
- Current supplier outlook: about 12x uplift in capacity
- InP optics market trajectory: $1.9B in 2025 to $22.75B by 2030
- Capacity targets by company: Lumentum near $9.0B; Coherent around $4.3B; Broadcom about $4.5B by 2030
- Analyst lens: Coherent and Lumentum highlighted for efficiency gains and transceiver market share gains
Risks and Timing
The road from capacity pledges to realized supply is long. Building more fabrication lines, refining yields, and integrating new tools can stretch over several years. If suppliers push the uplift beyond 12x, prices could stabilize sooner, potentially easing the current squeeze. If not, AI deployment windows may face delays or require alternative networking strategies to meet demand.
Conclusion: The Bottleneck Frontier in AI Hardware
The AI era is reshaping where bottlenecks appear. Nvidia’s bid to 20x InP laser capacity signals a bold bet that faster, denser networking will be a foundational element of next-generation AI. For investors, the decisive factor will be execution: translating capacity announcements into timely, scalable output that can sustain rapid AI deployment without eroding margins. If the optics ecosystem can deliver as promised, it could lay the groundwork for a more resilient and expansive AI hardware market through 2030 and beyond.
Discussion