Nvidia’s $95.2 Billion Supply Chain Bet Signals a New Era in AI Hardware

Nvidia's purchase obligations jump 90% to $95.2B as Vera Rubin platform targets 90% inference cost cuts, challenging hyperscaler custom chips amid $1T pipeline.

The numbers coming out of Nvidia’s supply chain are staggering — and they tell a story far bigger than any single earnings beat. The chipmaker’s purchase obligations have surged nearly 90% to $95.2 billion, up from a long-standing plateau of $28 billion to $30 billion, while inventories have swelled to $21.4 billion. This isn’t just restocking. It’s a deliberate, capital-intensive ramp for the Vera Rubin platform, the successor to Blackwell that promises to reshape the economics of artificial intelligence.

Vera Rubin, a fully co-developed system integrating six chip types — Vera CPU, Rubin GPU, NVLink-6 switch, ConnectX-9 SuperNIC, BlueField-4 data processor, and Spectrum-6 Ethernet switch — is designed to deliver five times the inference performance and 3.5 times the training performance of Blackwell. More critically for Nvidia’s competitive position, inference costs are expected to drop by 90% compared to the Blackwell GB200 processor. That’s a direct challenge to the custom chips being developed by Amazon, Google, and Microsoft, which have increasingly sought to reduce their dependence on Nvidia’s hardware.

Hyperscaler Demand Fuels a $1 Trillion Pipeline

The ramp comes as Nvidia’s biggest customers double down on AI infrastructure. Microsoft has penciled in roughly $190 billion in capital expenditures for calendar 2026, with the bulk flowing into CPUs and GPUs. Amazon is planning $200 billion this year, primarily directed at its cloud division and generative AI. Alphabet, too, has reported surging demand for compute capacity.

These commitments align with Nvidia CEO Jensen Huang’s projection that orders for Blackwell and Vera Rubin platforms will total $1 trillion across 2026 and 2027. The hyperscalers are already lining up: AWS, Google Cloud, Microsoft, and Oracle Cloud Infrastructure are expected to offer Vera Rubin-based instances starting in the second half of 2026.

The foundation is solid. In the fourth quarter of fiscal 2026, Nvidia posted record revenue of $68.1 billion — up 73% year over year. Net income jumped 94%, and free cash flow doubled to $34.9 billion. For the full fiscal year, revenue hit $215.9 billion, a 65% increase. The company returned $41.1 billion to shareholders through buybacks and dividends, with an additional $58.5 billion in authorized repurchases still available.

The Custom Chip Threat and Nvidia’s Strategic Response

Not everything is smooth sailing. The same hyperscalers that drive Nvidia’s revenue are also investing heavily in their own application-specific integrated circuits. Counterpoint Research expects shipments of such custom ASICs to triple by 2027 compared to 2024 levels.

Nvidia’s countermove is strategic. Through its NVLink Fusion technology, the company is integrating third-party chips — including Marvell’s custom processors — into its proprietary interconnect fabric. This ensures Nvidia remains relevant in data centers where its own GPUs are being replaced by in-house designs. On the other side, AMD, Intel, and Broadcom are rallying behind the open UALink standard as an alternative to NVLink.

Quantum Computing: A New Frontier

Away from the data center, Nvidia has entered a new arena. The company unveiled NVIDIA Ising, its first open-source family of quantum-AI models. These are designed to help researchers calibrate quantum processors, with error correction that Nvidia claims is 2.5 times faster and three times more accurate than classical approaches. The announcement sent quantum computing stocks sharply higher.

The May 20 Earnings Test

Nvidia shares are trading at around €169, roughly 7% below the 52-week high of €182 set on April 27. The relative strength index sits near 50, indicating neither overbought nor oversold conditions. Of 57 analysts covering the stock, 57 rate it a buy, with two neutral and one sell recommendation. Wall Street expects revenue growth of 72% for the current year.

The next major catalyst arrives on May 20, when Nvidia reports first-quarter results for fiscal 2027. The company’s own guidance calls for revenue of $78 billion, plus or minus 2%, excluding any contribution from China. The quarter ended April 26, so the numbers are already in hand. The question is how far Nvidia can exceed its own forecast — and whether the massive supply chain commitments are translating into accelerated revenue growth.