Nvidia and Micron earnings reframe the 2027 AI infrastructure trade

Nvidia's data center revenue of $75.2 billion in a single quarter now exceeds what AMD, Intel, and IBM each book from data centers in a full year.

Nvidia's data center revenue of $75.2 billion, up 92 percent year over year, and Micron's memory results reframe the AI infrastructure trade heading into 2027, when hyperscaler capital spending could reach $1.1 trillion.

"The buildout of AI factories, the largest infrastructure expansion in human history, is accelerating at extraordinary speed," Jensen Huang, chief executive officer at Nvidia, said.

Nvidia's total revenue reached $81.6 billion, up 85.2 percent year over year, with non-GAAP earnings per share of $1.87 topping the $1.77 consensus. Networking inside the data center segment surged 199 percent to $14.8 billion as NVLink and InfiniBand demand tripled. The company guided second-quarter revenue to roughly $91.0 billion at a 75.0 percent gross margin, excluding any China data center compute. Micron, the largest U.S. memory maker, rides the same wave: AI accelerators require high-bandwidth memory (HBM, the fast memory stacked beside GPUs), and its results reflect memory pricing and AI demand.

The four largest hyperscalers — Amazon, Microsoft, Alphabet, and Meta — collectively guide to roughly $745 billion in combined capital expenditures for 2026, up from $700 billion earlier in the year. Goldman Sachs estimated in a June note that 2027 hyperscaler capital expenditure alone could reach roughly $1.1 trillion in its base case and as much as $1.4 trillion in a bullish scenario.

Data Center Revenue Now Outpaces Full-Year Peers

Nvidia's single-quarter data center figure of $75.2 billion exceeds what AMD, Intel, and IBM each generate from data centers across a full year. AMD's data center revenue reached $6.7 billion in the quarter ended June 27, 2026, up 107 percent year over year, against $16.6 billion for all of 2025. Intel's Data Center and AI segment posted $6.3 billion in the second quarter, up 59 percent, its strongest growth in years. IBM's Infrastructure revenue, which includes its AI-relevant Power servers, fell 7 percent to $3.84 billion as mainframe sales declined.

The gap reflects Nvidia's roughly 80 percent share of the AI accelerator market by revenue, with an estimated $193.7 billion in data center sales in fiscal 2026. Broadcom, the second-largest AI semiconductor supplier, guided AI chip revenue to $16.0 billion for the quarter ending August 2, 2026, more than double year over year, and reiterated a target of more than $100 billion for fiscal 2027. Both companies rely on TSMC's advanced packaging, where CoWoS capacity remains a bottleneck for HBM-equipped accelerators.

Memory Demand Sets Up Micron's 2027

Micron's results, not yet fully disclosed, hinge on the same build-out. Every GPU and custom accelerator needs HBM, and memory content per AI server is rising faster than unit growth. Nvidia's visibility to $1 trillion in Blackwell and Vera Rubin sales from 2025 through calendar 2027, with Vera Rubin production shipments beginning later in 2026, underpins that demand.

Nvidia shares closed at $219.22 on August 5, 2026, up 17.68 percent year to date, trading at 38 times forward earnings of $8.26. Analysts are 95 percent bullish with a $302.83 consensus target. The setup for investors is a memory and compute cycle that stretches into 2027, with the risk that hyperscaler efficiency gains outpace demand growth and force a capex pullback. If AI services revenue fails to justify the outlay, the $1.1 trillion base case for 2027 spending — and the chip orders behind it — could be revised down.

This article is for informational purposes only and does not constitute investment advice.