Short answer: NVIDIA was reported to be in advanced talks to acquire Lepton AI on March 26, 2025, in a deal valued at several hundred million dollars. NVIDIA declined to comment at the time. Later reporting said the acquisition closed and that Lepton co-founders Yangqing Jia and Junjie Bai joined NVIDIA, but NVIDIA has not publicly announced the transaction or disclosed its terms.
The deal matters because Lepton operated at the intersection of GPU rental, AI inference, developer tooling and cloud infrastructure. Its reported acquisition also came shortly before NVIDIA launched DGX Cloud Lepton, a marketplace connecting developers with GPU capacity from multiple providers. The product connection is strategically apparent, but NVIDIA has not publicly explained how Lepton’s technology or team was integrated.
As an Amazon Associate I earn from qualifying purchases.
What was originally reported about NVIDIA and Lepton AI?
On March 26, 2025, TechCrunch reported that NVIDIA was in talks to acquire Lepton AI. The Information described the discussions as advanced and reported a potential price of several hundred million dollars.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesThe report came from an unnamed person close to the company. That distinction matters: “in talks,” “advanced talks” and “nearing a deal” describe negotiations, not a completed transaction. NVIDIA declined to comment, and the available public source record does not include a NVIDIA acquisition announcement or disclosed transaction filing.
#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
Later, The Information reported that the acquisition had closed. That report said the deal was worth hundreds of millions of dollars and that Lepton’s co-founders joined NVIDIA. However, NVIDIA has still not publicly confirmed the acquisition or released its price, structure or closing date. The most accurate description is therefore: the acquisition was reported in March 2025 and later reported as completed, but its official terms remain undisclosed.
What did Lepton AI do?
Lepton was a GPU-cloud and AI-infrastructure startup. It reportedly rented or resold access to servers equipped with NVIDIA GPUs and provided tools for developers building, training and running AI models.
Its offering included:
- GPU rental and access to cloud infrastructure;
- model training and inference;
- tools for deploying AI applications;
- a Python-oriented application-development layer; and
- infrastructure intended to simplify deployment across available GPU capacity.
The Information described Lepton as competing with GPU-cloud and inference providers such as Together AI and RunPod. It should not automatically be described as a company that owned a large data-center fleet. The available reporting indicates a cloud-access and infrastructure model, not clear ownership of all the underlying facilities.
TechCrunch reported that Lepton raised an $11 million seed round from CRV and Fusion Fund in May 2023.
Who founded Lepton?
The reported co-founders were Yangqing Jia and Junjie Bai. Jia is known for creating the Caffe deep-learning framework and for senior roles involving AI and cloud infrastructure. A 2025 conference program identified Jia as Lepton’s co-founder and CEO and described the company as a developer-centric platform for running AI applications at scale.
The Information reported that Jia and Bai had previously worked at Alibaba and Meta, including work connected with PyTorch, and later joined NVIDIA after the reported acquisition. Those biographical and employment details should be attributed to that reporting rather than treated as a separate NVIDIA confirmation.
Why would NVIDIA want Lepton?
Moving beyond chips
NVIDIA’s influence has historically centered on accelerated-computing hardware and the software ecosystem around it. A GPU-cloud business gives NVIDIA a more direct relationship with developers and end users instead of relying entirely on hyperscalers, server manufacturers and independent cloud operators.
The Information framed the reported deal as part of NVIDIA’s broader push into cloud and enterprise software, where it increasingly overlaps with companies such as Amazon, Google and Microsoft.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Capturing inference and usage revenue
AI infrastructure is shifting from model training alone toward recurring inference workloads. Training can require short-lived bursts of expensive compute, while production inference can create ongoing usage across applications and services.
A company that helps customers deploy and run models could therefore give NVIDIA a route toward usage-based infrastructure and software revenue. The available reporting does not establish Lepton’s revenue, margins, customer count or profitability, so the financial upside should be treated as strategic logic rather than a proven result.
Making NVIDIA GPUs easier to access
Lepton’s developer-oriented model fit NVIDIA’s goal of making its GPUs easier to obtain and use. The simpler it is for developers to find capacity, deploy containers and run NVIDIA-optimized software, the more deeply NVIDIA’s hardware and software stack can become embedded in AI projects.
Acquiring cloud and infrastructure expertise
The founders’ experience with large-scale AI and cloud systems may have been as important as Lepton’s customer-facing product. That is a plausible acquisition rationale, but NVIDIA has not publicly stated that talent acquisition was the reason for the deal.
How does DGX Cloud Lepton fit into the story?
In May 2025, shortly after the acquisition talks were reported, NVIDIA announced DGX Cloud Lepton. NVIDIA describes it as a global compute marketplace connecting developers with GPU capacity from multiple cloud providers.
The company said the platform was designed for training, fine-tuning, testing and inference. It also highlighted integration with NVIDIA software such as NIM, NeMo, Blueprints and Cloud Functions. Participating providers announced at launch included CoreWeave, Crusoe, Firmus, Foxconn, GMI Cloud, Lambda, Nebius, Nscale, SoftBank and Yotta Data Services.
In June 2025, NVIDIA announced a European expansion involving providers including Mistral AI, Nebius, Nscale, Firebird, Fluidstack, Hydra Host, Scaleway, Together AI, AWS and Microsoft Azure. NVIDIA also described integration with Hugging Face’s Training Cluster as a Service.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →The timing and product overlap strongly suggest a connection between Lepton and DGX Cloud Lepton. But NVIDIA’s public announcements do not explicitly say that DGX Cloud Lepton was created from the acquisition, nor do they document which technology, employees or customer relationships transferred. It is safer to describe the relationship as a likely strategic connection rather than confirmed corporate lineage.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
Verified timeline
| Date | What happened |
|---|---|
| May 2023 | TechCrunch reported that Lepton raised an $11 million seed round from CRV and Fusion Fund. |
| March 26, 2025 | The Information reported that NVIDIA was nearing a deal to acquire Lepton for several hundred million dollars. |
| March 26, 2025 | TechCrunch independently summarized the reported talks. NVIDIA declined to comment. |
| May 18–19, 2025 | NVIDIA announced DGX Cloud Lepton, a marketplace connecting developers with GPU capacity from cloud providers. |
| June 11, 2025 | NVIDIA announced a European expansion and additional provider integrations. |
| Before August 18, 2026 | The Information reported that the acquisition had closed and that both Lepton co-founders joined NVIDIA. |
| June 29, 2026 | A secondary report said Jia had left NVIDIA roughly 14 months after the acquisition. NVIDIA did not independently confirm that report in the available source set. |
What the deal means for NVIDIA’s cloud strategy
The reported acquisition reflects a broader attempt by NVIDIA to capture more of the AI-infrastructure stack:
- Hardware: GPUs and accelerated systems remain the foundation.
- Software: CUDA, NIM, NeMo and related tools help make NVIDIA hardware useful and sticky.
- Infrastructure access: Cloud marketplaces and hosted services make capacity easier to obtain.
- Developer distribution: Direct relationships with developers can influence which hardware and software they choose.
- Inference: Production model serving creates recurring demand after training is complete.
DGX Cloud Lepton is not described by NVIDIA as a conventional hyperscale public cloud. It is a marketplace and coordination layer spanning multiple providers. That approach can expand regional access without requiring NVIDIA to own every data center, but it also means the user experience may vary across providers.
Could the acquisition create channel conflict?
NVIDIA’s cloud partners are also major customers for its GPUs. A marketplace can benefit those partners by directing more demand toward their available capacity, especially when customers need a particular region or GPU generation.
Free tools Windows power users keep installed
One-click scans. No signup required.
At the same time, NVIDIA may gain more control over the customer relationship, software layer and purchasing workflow. Cloud providers could reasonably worry that NVIDIA might eventually favor its own services, influence pricing or compete more directly for enterprise workloads.
This conflict is not proof that partners will be displaced. A marketplace may instead broaden access to smaller regional providers and help them reach customers they could not acquire independently. The outcome depends on pricing, provider neutrality, capacity allocation and the contractual relationship between NVIDIA and its partners.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What should GPU-cloud customers consider?
The reported acquisition does not by itself guarantee cheaper GPUs, more capacity or identical service quality. Customers evaluating DGX Cloud Lepton or alternatives should examine:
- GPU availability: Confirm that the required GPU is actually available in the required region and account tier.
- Total cost: Include storage, networking, egress, idle instances, minimum commitments and support.
- Performance consistency: Different providers may offer different networking, storage, scheduling and maintenance behavior.
- Software compatibility: Check CUDA, drivers, containers, PyTorch, NIM, NeMo and framework versions.
- Data locality: Verify regional placement, sovereignty requirements and cross-border transfer rules.
- Reliability: Review service-level commitments, maintenance windows, replacement procedures and capacity guarantees.
- Portability: Test whether workloads can move to AWS, Azure, Google Cloud, CoreWeave, RunPod or another provider.
- Security: Check tenant isolation, private networking, identity integration and compliance certifications.
- Support: Distinguish community support from enterprise support and formal NVIDIA escalation.
Marketplace breadth versus consistency
A multi-provider marketplace may improve availability and geographic coverage, but “NVIDIA-powered” does not mean every provider offers identical uptime, networking, storage or support.
Convenience versus lock-in
NVIDIA’s integrated software stack can reduce deployment work. The trade-off is that applications built deeply around NVIDIA-specific services may be harder to move to AMD GPUs, Google TPUs, AWS Trainium or other platforms.
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Cheap GPU hours versus total cost
A low advertised GPU-hour rate can be outweighed by data egress, storage, idle time, premium networking or support fees. Production buyers should compare the cost of completing a workload, not just the hourly instance rate.
Rental flexibility versus guaranteed capacity
Short-term capacity may be suitable for experimentation, prototyping and burst workloads. Low-latency inference and other mission-critical services generally require predictable capacity, performance guarantees and a clear support contract.
How does Lepton compare with alternatives?
- RunPod: A developer-oriented option for GPU instances and serverless workloads. It may suit experimentation and smaller deployments, but buyers needing hyperscaler-native services or guaranteed enterprise capacity should examine its limitations.
- Together AI: Focused on open-model inference, fine-tuning, training and model-serving APIs. It is more model-service-oriented than a general-purpose multi-provider GPU marketplace.
- CoreWeave: A specialized GPU cloud suited to large-scale AI workloads and capacity commitments. It may be less appropriate for casual, low-volume experimentation.
- Lambda: Offers GPU cloud, workstations, servers and enterprise AI infrastructure through a more conventional provider model.
- Nebius: An AI-focused cloud with its own infrastructure rather than simply a marketplace layer.
- Hyperscalers: AWS, Microsoft Azure and Google Cloud offer broader ecosystems, procurement processes and enterprise controls, though pricing and capacity can be more complex.
These services are not interchangeable. The right choice depends on workload type, geography, capacity guarantees, software requirements, compliance and portability.
Recommended Free Tools
What remains unknown?
Even after the reported closing, several important details have not been publicly disclosed by NVIDIA:
- the exact purchase price;
- the transaction structure and closing date;
- Lepton’s revenue, valuation, customer count and profitability;
- which technical assets and employees transferred;
- how much of DGX Cloud Lepton came from Lepton’s technology;
- how existing Lepton customers were treated;
- employee retention beyond the reported founder appointments; and
- the long-term branding and product roadmap.
A later secondary report said Yangqing Jia left NVIDIA in 2026. That claim should be treated cautiously because it was not independently confirmed by NVIDIA in the available source set. It may be relevant to the acquisition’s aftermath, but it does not establish whether the deal succeeded or failed operationally.
Bottom line
NVIDIA’s reported Lepton AI acquisition should no longer be described simply as an unresolved March 2025 rumor. The Information later reported that the deal closed and that Lepton’s founders joined NVIDIA. However, because NVIDIA has not publicly confirmed the transaction or disclosed its terms, the price, structure and technical integration remain reported rather than officially documented.
The strategic significance is clearer than the corporate details. Lepton’s GPU-cloud and developer infrastructure fit NVIDIA’s effort to move beyond selling chips into cloud access, inference, orchestration and software services. DGX Cloud Lepton appears closely connected in timing and purpose, but customers should evaluate the marketplace on actual GPU availability, total cost, reliability, regional constraints and portability—not on the acquisition headline alone.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
