Nvidia is widely expected to post another exceptionally strong earnings report on Wednesday, but longer-term questions are emerging about whether the company can maintain its leadership in AI processors as the market evolves.
After several years of commanding a near-monopoly on the chips used to train advanced AI models, Nvidia now confronts a change in demand toward processors that perform inference - the task of running trained models, answering queries and executing tasks in real time. The inference segment is substantially larger than the training market, but it is also more competitive, attracting both established semiconductor rivals and technology companies building their own silicon.
Traditional competitors such as Intel and AMD have been developing processors tailored to the smaller, cost-sensitive workloads that characterize much of the inference market. At the same time, leading cloud and internet companies have taken steps to internalize chip design: Alphabet has become a prominent challenger with custom tensor processing units and has struck deals valued in the tens of billions of dollars for that technology, while Amazon’s chip business - including its Trainium processors - is gaining traction.
"It’s less so Nvidia versus TPUs, Nvidia versus AMD. I think it’s more: is the Nvidia ecosystem as dominant moving forward, as some of these new inference workloads start to proliferate," said John Belton, portfolio manager at Gabelli Funds, which holds Nvidia shares.
Market moves this year underscore the changing competitive landscape. Nvidia’s stock has risen about 19% so far this year, but that gain lags other players that have seen sharper advances - a roughly two-fold surge for AMD, Intel and Arm, and a 27% rise in Alphabet. In response to the shifting opportunity set, Nvidia introduced a new central processor and an AI system in March built on technology from Groq, an inference-focused startup it acquired. Those products, however, are not factored into the company’s projection of $1 trillion in cumulative sales from its Blackwell and Rubin platforms by the end of 2027.
Because those new chips lie outside the current Blackwell-Rubin sales forecast, investors will be watching closely for evidence of an additional revenue driver that can underpin the company’s long-term projections. Supply dynamics are another critical focus. Nvidia’s commitments to suppliers expanded materially - rising from $50.3 billion to $95.2 billion between the last two quarters of its most recent fiscal year - yet the company has so far largely avoided the memory-chip supply constraints that have affected chipmakers such as Qualcomm and Apple.
Analysts expect a sharp acceleration in revenue in the April quarter. LSEG data shows consensus estimates point to a 79% year-over-year increase in sales, the fastest pace of growth in over a year. Adjusted profit is forecast to have jumped 81.8% to $42.97 billion. That surge is attributed to heavy spending by major cloud and social-media customers, with estimates that Big Tech will invest more than $700 billion in AI this year, up from about $400 billion in 2025.
Nvidia’s leadership has signaled confidence in supply availability. CEO Jensen Huang has said the company has secured enough inventory to meet demand for several quarters, a statement that has helped alleviate some capacity concerns. Still, other potential constraints and demand-side risks remain.
One immediate bottleneck could be the physical infrastructure to deploy GPUs. "The customers just simply don’t have place to put the GPUs. They want to own as much as they can. They want to buy as much as they can, but they don’t really have the data centers to put them into," said Chaim Siegel, analyst at Elazar Advisors, pointing to the pace of data center buildouts as a limiter on near-term absorption of additional hardware.
Geopolitics adds another layer of uncertainty. Nvidia has not yet sold its H200 chips in China, where local alternatives are being promoted. The recent trip by Jensen Huang alongside U.S. President Donald Trump has increased hopes that progress could be made, but the situation remains unresolved.
Analysts also note margin risks. First-quarter gross margins are expected to total 74.5%, but that level could face pressure later in the year from rising memory and chip-packaging costs and from the ramp-up of Rubin-based chips.
Summary of the near-term picture
Nvidia should report strong quarterly results driven by sustained customer spending on AI, but the company faces a market transition toward inference workloads that is drawing competition from both legacy chipmakers and in-house silicon efforts by cloud providers. Investors will monitor revenue, supply commitments and evolving margin pressures as the company seeks pathways to sustain growth beyond its current Blackwell and Rubin projections.
Key considerations for markets and sectors
- Semiconductor sector: Shifting demand from training to inference changes product opportunity sets and competitive dynamics within chips and memory supply chains.
- Cloud and data center infrastructure: The pace of data center expansion will influence the ability of large customers to deploy additional GPUs and related hardware.
- Big Tech capital spending: Major cloud and social platforms remain central customers, with projected AI investment increases materially shaping demand.
Key points
- Nvidia is expected to post a very strong quarter but faces strategic pressure as AI use shifts from training to inference workloads.
- Competition is intensifying from Intel, AMD and in-house chips from companies such as Alphabet and Amazon, with Alphabet making deals worth tens of billions for its TPUs.
- Supply commitments have nearly doubled quarter-over-quarter to $95.2 billion, and investors will watch for signs of new growth engines outside the Blackwell and Rubin roadmap.
Risks and uncertainties
- Data center capacity - a slower-than-expected buildout could limit near-term demand for GPUs and related hardware, affecting the data center and cloud services sector.
- Geopolitical and market access concerns in China - Nvidia has not yet sold H200 chips there and local alternatives are being promoted.
- Margin pressure - rising memory and chip-packaging costs and Rubin ramp-up could compress margins later in the year, impacting semiconductor profitability.