Nvidia's AI Data Center Expansion Goes Far Beyond GPUs as Revenue Surges Toward $108 Billion

Nvidia's Revenue Machine Keeps Accelerating Despite Enormous Scale

Nvidia (NVDA) delivered another staggering quarter in fiscal Q2 2027 (ending July 26, 2026), posting revenue of $96.2 billion — a 106% jump compared to the same period a year earlier. With management now projecting approximately 70% year-over-year revenue growth for the full fiscal year 2028, the chipmaker's trajectory remains one of the most closely watched stories in the technology sector.

For the third quarter of fiscal 2027, Nvidia guided for revenue of $108 billion, plus or minus 2%, which would represent roughly 89% year-over-year growth at the midpoint — remarkable growth for a company already operating at this scale.

Expanding Beyond the GPU Playbook

One of the most significant developments in Nvidia's business model is how much ground it has covered outside its traditional graphics processing unit stronghold. Management estimates that the revenue opportunity per gigawatt of AI infrastructure has grown substantially with each hardware generation — from approximately $18 billion with Hopper GPUs to $25 billion with Blackwell systems, and now to $40 billion with the upcoming Vera Rubin architecture.

This expansion reflects Nvidia's deepening footprint across CPUs, networking equipment, and the broader hardware ecosystem needed to build end-to-end AI data centers.

Networking and Ethernet Revenue Surge

Data from IDC highlights how quickly Nvidia is gaining traction in adjacent markets. The company's data center Ethernet switching revenue surged 192.7% year over year to $2.1 billion in the first calendar quarter of 2026, giving it a 21.5% share of that market. Networking revenue also climbed 18% sequentially in Q2, with Spectrum-X Ethernet revenue more than doubling — up 2.6 times year over year.

CPU Business Building Momentum

Nvidia's server CPU ambitions are also materializing. Grace CPU revenue surpassed $5 billion on a trailing-12-month basis, and the next-generation Vera CPU had already entered full production by the end of Q2. Management sees a total addressable market of approximately $20 billion in server CPUs and preliminarily expects CPU revenue to more than double in fiscal 2028.

Competition Is Real, but the Market Is Expanding

Advanced Micro Devices (AMD) and major cloud hyperscalers are actively developing competing AI accelerators, which could pressure Nvidia's GPU market share over time. However, analysts note that a shrinking share of a rapidly growing market does not necessarily translate to shrinking revenue. The more relevant question for many investors may be how much of each AI data center Nvidia can monetize across all product lines — not just GPUs.

Supply Constraints, Not Demand, Are the Limiting Factor

Perhaps the most telling detail in Nvidia's guidance is that management characterized its fiscal 2028 revenue outlook as supply-constrained rather than demand-constrained — meaning customer forecasts actually point to even faster growth if supply were available. Nvidia expects this bottleneck to persist through the end of fiscal 2028.

To secure capacity, Nvidia has made substantial commitments to memory and manufacturing partners. Supply and capacity commitments ballooned from $119 billion at the end of Q1 to $279 billion at the end of Q2 fiscal 2027 — a figure that underscores both the scale of ambition and the risk involved. Some of these arrangements can be modified or canceled before firm orders are placed, but if AI infrastructure spending were to slow unexpectedly, the company could find itself overextended on supply commitments.

Vera Rubin Ramp Adds Execution Complexity

Nvidia began production shipments of Vera Rubin systems in August 2026, with purchase orders already secured from every major hyperscaler, AI cloud provider, and system manufacturer. Management expects Vera Rubin to contribute around 20% of data center revenue in Q3 — an aggressive ramp that overlaps with continued strong demand for Blackwell systems.

Product transitions historically carry execution risk in the semiconductor industry, and managing two high-demand generations simultaneously adds operational complexity. So far, Nvidia's scale does not appear to have slowed its ability to execute.

Valuation in Context

As of August 26, 2026, Nvidia was trading at approximately 23.2 times analysts' consensus fiscal 2027 adjusted earnings per share estimate of $9.05, and around 16 times the fiscal 2028 EPS estimate of $13.13. Other notable semiconductor names analysts often discuss alongside Nvidia include Broadcom (AVGO) and Taiwan Semiconductor Manufacturing (TSM).

Whether Nvidia's valuation is considered stretched or justified depends heavily on how one views the durability of AI infrastructure spending — and whether the company can continue converting its dominant market position into earnings growth over the next several years.

What to Watch

Key metrics worth monitoring in coming quarters include the pace of Vera Rubin adoption, growth in CPU and networking revenue relative to GPU revenue, any signals of AI capex moderation from hyperscalers, and how Nvidia manages its rapidly growing supply commitment book. The interplay between these factors will shape the company's financial profile well into 2027 and beyond.