The chip giant consolidates its ‘all-in-one’ strategy to offer artificial intelligence as infrastructure and service.
NVIDIA continues to evolve. Once the global benchmark for AI hardware, it now takes another step towards full dominance of the cloud ecosystem with the acquisition of Lepton AI, a startup specializing in the efficient rental of servers for AI-based workloads.
As reported by The Information, although an official announcement has yet to be made, the deal is practically closed and aligns with Jensen Huang’s strategic goal: to transform NVIDIA into a comprehensive provider of AI as Infrastructure (IaaI) and AI as a Service (AIaaS).
Infrastructure and services under one roof
The proposal from Lepton AI caught NVIDIA’s attention due to its agile and cost-effective operational model. By specializing in the efficient deployment of cloud resources for AI, it has managed to optimize costs without sacrificing performance. Just what NVIDIA needs to expand its cloud offerings beyond its flagship product, DGX Cloud.
With this acquisition, NVIDIA will be able to integrate Lepton’s infrastructure into its service offerings for:
- Companies looking to outsource their AI training and inference.
- Startups seeking scalability without building from scratch.
- Projects requiring access to optimized clusters without capital investments.
Huang’s vision: All-in-one AI
The strategy becomes clear after what was seen at GTC 2025: NVIDIA no longer just designs chips. It also builds the infrastructure, provides the frameworks, enables the software, and with Lepton AI, operates the service.
“Design, cluster, data center, and service. All under the NVIDIA brand. For the customer, frictionless AI,” industry analysts point out.
Far from being a tactical acquisition, Lepton AI is another piece of the plan to offer a fully controlled vertical platform, where the company can provide everything from hardware to operational support and AI management software.
AI as a service: the new market
NVIDIA is no longer only competing in silicon. With the purchase of Lepton AI, it dives straight into the battle of AI-as-a-Service (AIaaS), directly facing hyperscalers like AWS, Microsoft Azure, and Google Cloud.
The difference: the entire stack—chips, servers, frameworks, and connectivity—is optimized end-to-end. This translates to greater energy efficiency, lower latency, and better performance per watt, which is crucial in today’s context of accelerated demand and rising costs.
Conclusion: a more cloud-focused NVIDIA than ever
This move reinforces NVIDIA’s vision as a dominant player in the future of cloud AI. It is no longer just the supplier of GPUs that power large models; it is beginning to control how they are trained, where they are deployed, and how they are offered as a service.
With Lepton AI, NVIDIA positions itself as the “AWS of AI”, capable of competing in a market that is rapidly migrating from physical infrastructure to managed services. And it does so with a key advantage: no one else controls the hardware like they do.