Version 2

NVIDIA's acquisition of LeptonAI has effectively unwound: founder and CEO Yangqing Jia departed roughly one year after the deal closed, the DGX Lepton product failed commercially, and a public commitment to open-source the platform's core software by 2026 has gone unfulfilled [2][1]. NVIDIA reportedly paid ~$700M for LeptonAI [2], whose founder co-created Caffe, ONNX, and PyTorch [1]. The story broke via SemiAnalysis on June 28, 2026, and has since been picked up by secondary outlets including TechTimes, KuCoin, and BigGo Finance [11][10][9], though no new primary sources have emerged. Neither NVIDIA nor Jia has issued a public statement.

If NVIDIA's culture systematically stifles acquired ML platform companies, that is a meaningful constraint on its software-layer ambitions. The broken open-source commitment is notable given that the ML ecosystem largely runs on open tools — and the CEO who left helped build several of them.

NVIDIA acquired LeptonAI — founded by Yangqing Jia, co-creator of Caffe, ONNX, and PyTorch — for approximately $700M, with a stated commitment to open-source the platform's core software by 2026 [1][2]. DGX Cloud Lepton was positioned as a marketplace connecting developers to GPU compute across cloud providers [3][4]. As of mid-2026, the open-source commitment has not been met, the product has failed commercially, and Jia has departed roughly one year after the deal closed [2][1].

The only substantive public account comes from SemiAnalysis, which reported the outcome on June 28, 2026, and frames it as the predictable result of NVIDIA's product management culture. The firm had been cautiously optimistic about Lepton's independent direction as recently as March 2025 and considered Lepton a credible competitive threat by mid-2025 [5][6]. Its current position is that NVIDIA's culture consistently prevents acquired ML platform companies from reaching their potential, and that Lepton is one instance of a broader pattern [6].

On why Jia left, SemiAnalysis offers one hypothesis: Jensen Huang reversed an earlier approval to open-source Lepton's core platform, creating a strategic disagreement large enough to prompt the founder's exit despite standard multi-year vesting [7]. This remains unconfirmed. SemiAnalysis notes that Jia's track record made the original open-source commitment more credible than it might otherwise have been, which makes its non-fulfillment more pointed [1].

SemiAnalysis adds a broader market observation: agentic coding tools can now produce infrastructure software that platforms like Lepton were meant to supply, citing a SLURM-compatible replacement built by @AnushElangovan as an example [8]. The implication is that the market window for an NVIDIA-controlled ML platform may have narrowed regardless of the cultural issues. Secondary outlets have amplified the SemiAnalysis narrative [9][10][11], but no new primary reporting, direct statements from Jia, or NVIDIA responses have appeared.