NVIDIA invests up to $3B in Lancium and joins $500B Wall Street AI infrastructure deal

NVIDIA is deepening its move from chip supplier to infrastructure financier, committing up to $3 billion to Lancium — the developer behind the Texas Stargate data center campus — with an initial $2 billion buying roughly a 20% stake. The investment ties NVIDIA directly to the buildout of the compute capacity its own GPUs will fill, vertically aligning silicon supply with data-center demand.

Separately and more strikingly, NVIDIA is joining a consortium of Wall Street heavyweights — Apollo, Blackstone, BlackRock's Global Infrastructure Partners, Brookfield, Goldman Sachs, and KKR — on a potential $500 billion AI infrastructure funding package. The scale signals that AI compute buildout has become a mainstream institutional asset class, with the largest private-capital players treating data centers and power as core infrastructure bets.

The deals fit a clear theme-of-the-week: compute-investment megadeals. They land alongside Meta's $1 billion data-center community fund and Microsoft's disclosure of a $678 billion commercial cloud backlog with 43% Azure growth — all pointing to an industry pouring unprecedented capital into physical AI infrastructure.

Skeptics see circularity risk: NVIDIA financing the customers who buy its chips inflates demand signals and concentrates systemic exposure if the AI capex cycle turns. The $500B figure is also described as a 'potential' package, not a closed deal. What to watch: whether the Wall Street consortium formalizes, how much of Lancium's Stargate capacity is pre-committed, and whether regulators scrutinize NVIDIA's growing web of customer-and-supplier equity stakes.