NVIDIA、Lumentumに20億ドルを投資してAI向け光学と米国製造を加速
NVIDIA's $2B Bet on Lumentum
NVIDIA announced a multiyear strategic agreement with Lumentum that includes a $2 billion investment to expand U.S.-based manufacturing capacity and deepen R&D collaboration in data center optics [1]. The deal also comprises a multibillion-dollar purchase commitment for advanced laser components and modules that underpin next-generation optical interconnects for AI datacenters [1].
Why optics matter for AI scale
Silicon photonics and high-performance lasers are becoming critical to scale AI infrastructure without unsustainable power and space costs. By anchoring production in the U.S., the partnership aims to reduce supply-chain friction for silicon photonics assemblies and improve the energy efficiency and resiliency of large-scale AI networks [1]. As Jensen Huang observed in NVIDIA’s announcement, AI has reshaped computing and is driving an unprecedented infrastructure buildout—this investment is a direct play on that shift [1].
Timing and industry context
Doppler VPNでプライバシーを保護
3日間無料体験。登録不要。ログなし。
The announcement arrived alongside Mobile World Congress activity in Barcelona, where optics and network modernisation have been focal topics for carriers and cloud providers alike [2]. The move signals a concerted effort by a major AI platform provider to vertically integrate more of the optical stack, rather than relying solely on commodity suppliers.
Practical implications
For companies building AI services, a steadier supply of advanced optical components could lower deployment friction and shorten lead times for hyperscale interconnect upgrades. For policymakers, the deal highlights how strategic industrial investment—paired with onshore manufacturing—can be used to bolster resilience in critical tech supply chains.
Investors and procurement teams should watch timelines for the new U.S. capacity and the cadence of the purchase commitments: those details will determine when optical-priced improvements and latency gains flow into real-world datacenters. For engineers, the partnership accelerates the timeline for integrating denser, more energy-efficient optical links into AI systems.
Sources: