NVIDIA Q2 FY2027 Earnings: Revenue Breakdown by Segment and Growth Drivers

NVIDIA (NVDA) reported quarterly revenue of $96 billion in Q2 FY2027, more than double its result from the year-ago period, as enterprise and government spending on AI infrastructure accelerated alongside continued hyperscale buildout. The data center segment drove the bulk of the result, generating $89 billion in revenue, up 18% sequentially. The NVIDIA Q2 FY2027 earnings podcast on Beta Finch covers the full call, including segment detail, management commentary, and analyst Q&A.

The quarter's most structurally notable development was the near-equal split within that data center figure. Hyperscale customers, the large cloud providers constructing AI infrastructure at scale, accounted for $49 billion, up 13% sequentially. The remaining $40 billion came from NVIDIA's ACIE segment, which covers NeoClouds, enterprise AI deployments, and sovereign AI programs.

Hyperscale vs. ACIE: A Narrowing Gap

ACIE revenue rose 25% sequentially and 138% year-over-year in Q2 FY2027, outpacing the hyperscale cohort on both dimensions. CFO Colette Kress noted on the earnings call that ACIE now represents roughly half of NVIDIA's data center business. A year ago the segment was a fraction of the hyperscale bucket; it has since closed the gap through a combination of enterprise on-premise buildout and accelerating national AI programs.

Sovereign AI revenue tripled year-over-year, reflecting procurement waves from governments pursuing domestic compute capacity across Asia, the Middle East, and Europe. These programs tend to fund multi-year deployments rather than quarterly purchasing cycles, introducing a structural revenue visibility component that differs from spot hyperscale demand. The scale of sovereign commitments contributed meaningfully to ACIE's 138% year-over-year growth rate and the segment's emergence as a roughly equal counterpart to hyperscale within the data center business.

Supply-Constrained Growth

NVIDIA guided Q3 FY2027 revenue to $108 billion, plus or minus 2%, representing sequential growth of roughly 12.5% from Q2. For full fiscal year 2028, the company projected revenue growth of approximately 70%. CEO Jensen Huang characterized that figure explicitly as supply-constrained, stating that actual demand is tracking closer to 100% annual growth.

The roughly 30-percentage-point gap between guided growth and demand-implied growth reflects manufacturing capacity and supply chain limitations rather than any moderation in end-market conditions. NVIDIA's ability to narrow that gap depends on supply additions from its manufacturing partners over the next several quarters. Prior-quarter context on NVIDIA's ramp trajectory is available on the semiconductors group page.

Gross Margin Headwinds and the Recovery Path

Gross margin faces a defined near-term compression. NVIDIA guided Q3 gross margin to 74%, with management projecting a further trough of 71-72% in Q4 FY2027. The primary driver is elevated DRAM pricing, which raises the cost structure of current GPU product configurations and compresses the spread between revenue per unit and manufacturing cost.

The margin pressure is closely tied to Blackwell-generation shipments, which carry higher memory content per unit than the Hopper architecture they replaced. Management expects the trough to be temporary. As DRAM spot prices normalize and product mix evolves, gross margin is guided to recover to 72-73% in fiscal year 2028, returning closer to the company's prior operating range.

Vera Rubin Architecture and the Revenue-Per-Gigawatt Framework

NVIDIA's next GPU architecture, Vera Rubin, was presented as a step-function improvement in compute density and operating economics. Management cited 30x higher throughput per megawatt and 35x lower token cost versus the Blackwell Ultra generation. The efficiency gains have direct implications for how enterprise and sovereign customers plan infrastructure expansions, as cost-per-token delivered to end users falls sharply with each generation.

NVIDIA framed the generational progression through a revenue-per-gigawatt metric, measuring the monetization potential of each installed gigawatt of compute capacity. Hopper-generation deployments represent approximately $18 billion in revenue opportunity per gigawatt. Blackwell raises that figure to roughly $25 billion per gigawatt. Vera Rubin is projected to reach approximately $40 billion per gigawatt, representing a 60% step-up from Blackwell and more than double the Hopper baseline.

The per-gigawatt progression reflects both higher compute density per rack and declining cost-per-token delivered at scale. Management presented this framework as a lens on the economics of platform upgrades: each successive generation extracts materially more revenue from the same physical datacenter footprint, a dynamic that management cited as a driver of planned refresh cycles across the installed base.

Customer Commitments and Forward Guidance

Amazon's deployment plans illustrate the scale of hyperscale commitments over the forward period. The company is deploying an additional 2 million GPUs through fiscal 2029, alongside Vera CPUs paired with Rubin GPUs. The partnership extends NVIDIA's hardware revenue visibility into the next architecture generation across one of its largest cloud customers.

The Q3 guidance of $108 billion and the approximately 70% FY2028 growth projection define the near-term revenue trajectory. The key variables through that period are the pace of DRAM price normalization, which shapes the gross margin recovery path; the Vera Rubin production ramp schedule; and whether the ACIE segment sustains its current growth rate differential versus the hyperscale cohort.

• Total Q2 FY2027 revenue: $96 billion, more than double year-over-year
• Data center revenue: $89 billion, up 18% sequentially
• Hyperscale revenue: $49 billion, up 13% sequentially
• ACIE segment revenue: $40 billion, up 25% sequentially and 138% year-over-year; now roughly half of data center, per CFO Colette Kress
• Sovereign AI revenue: up approximately 3x year-over-year
• Q3 FY2027 revenue guidance: $108 billion, plus or minus 2%
• FY2028 revenue growth: approximately 70% guided (supply-constrained); actual demand closer to 100% per Jensen Huang
• Gross margin path: 74% in Q3, trough of 71-72% in Q4, recovery to 72-73% in FY2028
• Vera Rubin vs. Blackwell Ultra: 30x throughput per megawatt, 35x lower token cost; approximately $40 billion revenue opportunity per gigawatt