Half a year spent on throwing $18 billion! NVIDIA's investment map reveals the trend of AI investment - optical interconnection and computing power rental.

Attention, in the past six months, NVIDIA Corporation (NVDA.US) has invested a total of at least $18 billion in strategic investments in listed companies. The tech giant is continuing to strengthen long-term partnerships and accelerate the AI computing power infrastructure update iteration process, while maintaining the supply-demand balance downstream. A recent investment was announced today, NVIDIA Corporation invested $2 billion in Marvell Technology, Inc. (MRVL.US). The two companies unveiled a strategic partnership aimed at connecting Marvell to NVIDIA Corporation's AI factories and AI-RAN ecosystem through NVLinkFusion, highlighting NVIDIA Corporation's desire to work with Marvell Technology, Inc. to accelerate data center optical interconnection technology. Earlier this month, NVIDIA Corporation disclosed a $2 billion investment in Nebius (NBIS.US), a leading AI computing leasing company headquartered in the Netherlands. This partnership will help Nebius deploy over 5 gigawatts (GW) of NVIDIA Corporation systems by the end of 2030. Prior to this, NVIDIA Corporation also invested $2 billion each in optical networking companies Lumentum (LITE.US) and Coherent (COHR.US). These investments will help the companies, led by Huang Renxun, ensure access to advanced laser and optical networking products in the future. Looking at the overall development trend of data center optical interconnection at OFC 2026, the industry's main focus is very clear: the bottleneck of large-scale AI data centers is shifting from single-chip computing power to "cluster-level interconnection bandwidth, energy consumption, and scalability." Lumentum is a maker of critical components in optical communications (laser, optical device suppliers) and also ships some optical module products and subsystems. However, its core technological advantage lies in its ability to exclusively build the most basic light sources and high-speed optical devices platform, which are widely integrated into various optical modules and optical communication systems in large AI data centers. Both TPU and GPU computing clusters have a common core feature: they require high bandwidth, low latency, and high energy efficiency internal interconnects in data centers. Traditional copper cables or electronic switching solutions cannot meet the demands when scaled to thousands or even tens of thousands of chips, as the power consumption and heat dissipation increase explosively; optical interconnection technologies (mainly including Co-packaged Optics CPO, silicon photonics switches, Optical Circuit Switches OCS) can replace electrical signals with optical signals, significantly improving bandwidth density and energy efficiency, reducing latency and power consumption in large-scale AI training/inference networks. This demand for higher optical interconnection technology capacity is common to both GPU and TPU clusters. For example, NVIDIA Corporation's Silicon Photon Network Switches such as Spectrum-X/Quantum-X explicitly integrate laser and photon technologies to improve power efficiency and network capacity, and Lumentum's high-performance laser and optical components are an essential part of these switches. With access to capital, production capacity, and research and development binding, Lumentum is likely to become a primary beneficiary, especially in core optical interconnects such as UHP lasers/ELS/CPO/OCS, in the future data center AI optical interconnection wave. Lumentum is highly likely to become a "core AI arms dealer" in the future data center AI optical interconnection wave, especially for NVIDIA Corporation's AI GPU computing chain. In January of this year, NVIDIA Corporation invested $2 billion in another large AI computing leasing platform, CoreWeave (CRWV.US), to help this AI large-scale computing service provider accelerate the construction of a 5 gigawatt AI factory by 2030. At the end of 2025, NVIDIA Corporation purchased a large number of shares from Intel Corporation (INTC.US) for $5 billion. "This historic collaboration will tightly integrate NVIDIA Corporation's AI and accelerated computing stack with Intel Corporation's CPU and vast x86 ecosystem the convergence of two world-class platforms," Huang Renxun said. "We will jointly expand the ecosystem and lay the foundation for the next era of computing." At the end of 2025, NVIDIA Corporation also invested $2 billion in Synopsys, Inc. (SNPS.US) stocks, with the aim of integrating NVIDIA Corporation's AI and accelerated computing advantages with Synopsys, Inc's engineering solutions. This is expected to enable the research and development team to design, simulate, and verify intelligent products with higher precision, speed, and lower cost. In October of last year, NVIDIA Corporation announced a $1 billion investment in Nokia Oyj Sponsored ADR (NOK.US) to integrate their technologies, helping both parties become leaders in the AI-driven network field. NVIDIA Corporation has also recently established partnerships with a wide range of energy companies, such as SLB (SLB.US), AES (AES.US), NextEra Energy (NEE.US), and Vistra (VST.US), to ensure the power supply for its AI factories. Although not a listed company, NVIDIA Corporation's largest single investment during this period was a $30 billion investment in OpenAI confirmed in late February. NVIDIA Corporation has made significant investments in dozens of AI-related startups, including core competitors of OpenAI such as Anthropic and xAI.