Nvidia Data Center Growth in 2026

Nvidia (NVDA) reported $89.0 billion in data center revenue for the quarter ended July 26, 2026, up 117% from the year earlier, according to the company’s Q2 fiscal 2027 press release. That single quarter exceeds the segment’s entire fiscal 2024 total of $47.5 billion. Total company revenue reached $96.221 billion, up 106% year over year, and GAAP net income was $59.688 billion, up 126%.

Data center server racks powering AI compute

Key Takeaways

The Q2 Fiscal 2027 Data Center Numbers

The data center segment made up about 92% of Nvidia’s $96.221 billion quarterly revenue, up from roughly 88% a year earlier. The segment grew 18% sequentially from $75.2 billion in the prior quarter. GAAP operating income reached $63.734 billion, up 124% year over year, and GAAP diluted earnings per share were $2.46 compared to $1.08 a year earlier.

Gross margin remained at 75.0% on both a GAAP and non-GAAP basis, rising 2.6 percentage points from 72.4% a year earlier. This reflects Nvidia’s pricing strength: a hardware company maintaining software-like margins at a scale where most semiconductor firms experience margin pressure. The company’s guidance for Q3 fiscal 2027 projects gross margin to decline to 74.0%, plus or minus 50 basis points, due to rising memory costs.

Nvidia returned about $26.0 billion to shareholders during the quarter through buybacks and dividends, leaving approximately $99.0 billion under its repurchase authorization. It also declared a quarterly cash dividend of $0.25 per share, payable October 1, 2026, to shareholders of record on September 10, 2026.

Custom Silicon and the Margin Question

The fastest-growing competitive threat to Nvidia is not a rival GPU. TrendForce data cited by CommandLinux shows custom ASIC shipments growing 44.6% in 2026, nearly triple the 16.1% rate for merchant GPUs. ASIC-based servers are on track for 27.8% of accelerator server shipments in 2026, the highest share since 2023. Broadcom (AVGO), the leading custom chip architect, has a $73 billion AI backlog.

Custom silicon from Google (GOOGL), Amazon (AMZN), Microsoft (MSFT), and Meta (META) together accounts for an estimated 12% to 18% of accelerator revenue. These chips are internal cost items for their owners rather than merchant products, so they reduce Nvidia’s addressable market in specific workload categories instead of competing for the same purchase order. The displacement is concentrated in inference, where workloads are predictable and volumes justify custom design. Training remains more firmly with Nvidia because CUDA’s flexibility keeps diverse enterprise workloads on merchant GPUs.

Nvidia’s approach is to sell the entire rack rather than just the chip. The Vera Rubin platform is ramping into full production with racks running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius. The company also announced Spectrum-6 switch systems supporting both pluggable and co-packaged optics as part of the platform. Selling networking, cooling, and software alongside the accelerator increases switching costs beyond a single component comparison.

The margin risk lies in memory. Nvidia reportedly reduced memory configuration in its Rubin platform as component availability tightened and prices increased, and management guided Q3 gross margin down to 74.0% because of rising memory costs. Nvidia’s supply commitments more than doubled to $279 billion, a figure that secures years of memory supply but also becomes a fixed obligation if AI demand slows.

Stock Performance and Analyst Outlook

NVDA closed at $224.58 on September 24, 2026, down 0.41% on the session. The stock has lagged AMD significantly on a year-to-date basis: AMD is up roughly 187% in 2026 while Nvidia has gained about 22%, according to Yahoo Finance’s comparison of the two stocks. Nvidia’s market capitalization was about $5.45 trillion as of September 23, compared with AMD’s newly crossed $1 trillion.

The valuation gap runs the other way. AMD trades at roughly 40 times expected 2027 earnings while Nvidia trades at about 15 times its own, per the same Yahoo Finance analysis. That difference reflects the market’s expectation that AMD’s growth rate is less certain and that Nvidia’s earnings base is already very large.

Guidance is the main forward-looking indicator. Nvidia’s Q3 fiscal 2027 outlook of $108.0 billion assumes no data center compute revenue from China. The company’s fiscal 2028 revenue growth forecast of 70% exceeded analyst estimates, as CNBC reported in its live coverage of the August 26 earnings release. That forecast is management’s own projection, not an independently verified result, and it depends on hyperscaler capital spending continuing at current levels.

The China assumption is the largest single variable. Nvidia’s outlook excludes a market that historically contributed more than 20% of revenue. Any change in export policy that reopens Chinese data center compute sales would add revenue above the guided figure; continued restrictions keep that demand flowing to Huawei’s Ascend line and domestic alternatives.

Nvidia Versus Its Competitors: A Side-by-Side Comparison

Metric	Nvidia	AMD	Intel
Data center segment revenue (most recent reported quarter)	$89.0 billion (Q2 FY2027, ended July 26, 2026)	$5.8 billion (quarter ended March 2026)	$5.1 billion (Data Center and AI segment)

Conditions for the Thesis

Three factors support Nvidia’s current valuation. First, hyperscaler capital spending must continue growing. The demand side is visible in backlog figures: Google Cloud backlog reached $514 billion, Microsoft’s commercial remaining performance obligations reached $678 billion, and AWS backlog reached about $496 billion, as we detailed in our analysis of hyperscaler capex and AI infrastructure. Those commitments are contracted demand, but they still require delivered capacity and customer consumption to convert into Nvidia revenue.

Second, gross margin must stay near 75%. Memory cost inflation is the specific pressure point, and the Q3 guide of 74.0% is the first decline. If HBM and DRAM prices keep rising while Nvidia cannot fully pass them through, the margin that funds its R&D and TSMC capacity priority shrinks.

Third, the $279 billion in supply commitments must be matched by demand. That figure secures the memory and packaging capacity Nvidia needs to ship, but it also becomes a fixed cost if the AI buildout slows. The commitment more than doubled from the prior level, so the company has taken on more inventory risk in exchange for supply certainty.

Nvidia’s competitive position in 2026 is strong but facing more competition. It controls roughly three quarters of accelerator revenue, ships its newest platform while AMD’s rack-scale systems are still ramping, and holds priority access to TSMC’s CoWoS packaging capacity. AMD has credible hyperscaler contracts and a lower valuation, and custom ASICs are taking inference share at the companies that buy the most accelerators. The market is growing fast enough that both can expand, but Nvidia’s percentage share will keep shrinking as long as custom silicon grows nearly three times faster than merchant GPUs.

Sources and References

Sources cited while researching and writing this article: