As the artificial intelligence wave continues to surge, semiconductor giant Nvidia is quietly executing an unprecedented capital expansion. Led by CEO Jensen Huang, the silicon powerhouse is leveraging a massive $90 billion deal spree encompassing investments, acquisitions, and strategic partnerships to bankroll the global AI boom. This aggressive campaign not only solidifies Nvidia’s dominance in hardware but also extends its reach deep into software, large language models (LLMs), embodied AI, and AI agents.
In this high-stakes capital game, Nvidia acts as a kingmaker. Through its venture arm NVentures and corporate development teams, the company has backed dozens of high-profile AI startups over the past year, including Cohere, Mistral AI, Adept, and the specialized cloud provider CoreWeave. For these startups, securing an investment from Nvidia is less about the capital and more about gaining VIP access to highly coveted H100 and Blackwell GPU allocations. This "chips-for-equity" dynamic has created an incredibly resilient moat in Silicon Valley.
Market analysts point out that Nvidia's investment thesis creates a powerful closed-loop feedback system. By funding promising AI startups, Nvidia ensures that these companies route their capital back into purchasing Nvidia chips or renting GPU compute from Nvidia-backed cloud partners. While this dynamic has fueled blockbusting revenue growth for the chipmaker, it has also drawn intense scrutiny from antitrust regulators concerned about market monopolization and "round-tripping" financial dynamics.
However, Jensen Huang’s ultimate vision extends far beyond selling hardware. As generative AI transitions from static chat interfaces to autonomous AI Agents capable of executing complex workflows, Nvidia is rapidly building out its software and agentic frameworks. From Project GR00T for humanoid robotics to Nvidia NIM (Nvidia Inference Microservices) for enterprise agent deployment, the company is positioning itself as the indispensable nervous system of the upcoming agentic era.
[AgentUpdate Depth Analysis] Nvidia’s $90 billion deal spree is a pre-emptive strike to capture the "post-GPU" AI Agent ecosystem. Unlike Microsoft’s tight vertical integration with OpenAI or Google’s proprietary TPU stack, Nvidia is pursuing an infrastructure-level "swarm strategy." By funding both foundational compute networks and upstream Agent orchestrators, Nvidia is transitioning GPUs into the mandatory industrial infrastructure of the agentic era. For developers, this means future AI Agents will be structurally dependent on CUDA and Nvidia NIM microservices. This capital-driven strategy shifts the Agent competition from software layer down to physical compute and proprietary protocols, posing a long-term ecosystem challenge to open-source agent communities and rival silicon vendors.
