NVIDIA has deployed approximately $90 billion in partnerships and investments over the past 16 month...
The AMW Read
NVIDIA's $90B investment blitz updates the player map (04.§2) with a new capital-allocation model, triggers cross.§D explicitly ($90B > $500M), and triggers cross.§E due to explicit regulatory scrutiny in US/EU/UK.
NVIDIA has deployed approximately $90 billion in partnerships and investments over the past 16 months, spanning more than 145 companies including AI model developers, cloud providers, and AI infrastructure suppliers. The aggressive push, detailed by Financial Times and PitchBook data, includes $47 billion from January 2024 to January 2025, followed by an additional $43 billion. Key deals include $3.4 billion in GPU capacity agreements with neocloud provider Irent, $2 billion investments in Marvell and photonics firms Coherent and Lumentum, and $3.2 billion to Corning. The strategy intertwines investment with commercial partnerships, often tied to NVIDIA's NVLink interconnect and its open-source Nemotron model family, deepening ecosystem lock-in.
Why it matters: NVIDIA is transforming from a chip supplier into the central capital allocator of the AI industry, deploying cash equal to ~40% of its operating cash flow — far surpassing Google's 6%. This mirrors the hyperscaler-distribution pattern but at the infrastructure layer: by funding compute-hungry startups and tying them to CUDA and Nemotron, NVIDIA creates a self-reinforcing moat. The approach also echoes the acqui-licensing pattern seen in its $20 billion Grok deal, where technology and talent are internalized to bolster its software stack. However, regulators in the US, EU, and UK are scrutinizing these moves, marking an open debate about whether NVIDIA's ecosystem play constitutes anti-competitive behavior.
Grounded expert take: The scale of NVIDIA's capital deployment — $90 billion in 16 months — signals a structural shift in AI infrastructure economics. By acting as both investor and supplier, NVIDIA is effectively verticalizing the AI cloud market, preempting hyperscaler competitors like Google and Amazon. The risk is financial overextension and regulatory blowback: the article notes that CoreWeave's existence depends on NVIDIA's support, highlighting the fragility of this model. For competitors, the message is clear: winning against NVIDIA requires not just better silicon but a competing capital-allocation machine.
#NVIDIA #AIInfrastructure #EcosystemMoat #RegulatoryScrutiny #CapitalCycles #HyperscalerDistribution
