Lepton AI
Founded Year
2023Stage
Acquired | AcquiredTotal Raised
$11MAbout Lepton AI
Lepton AI focuses on AI model deployment and cloud-native infrastructure within the AI industry. The company provides services including reliability in deployment, development environments, and support for job processing and collaborative workflows. Lepton AI serves sectors that require AI model deployment and inference. It was founded in 2023 and is based in Cupertino, California. In April 2025, Lepton AI was acquired by NVIDIA.
Loading...
Loading...
Expert Collections containing Lepton AI
Expert Collections are analyst-curated lists that highlight the companies you need to know in the most important technology spaces.
Lepton AI is included in 1 Expert Collection, including Artificial Intelligence.
Artificial Intelligence
7,221 items
Latest Lepton AI News
Apr 9, 2025
来源：DoNews 4月8日消息，据 The Information 援引知情人士消息，英伟达已完成对 AI 大牛、阿里巴巴前技术副总裁贾扬清创企 Lepton AI 的收购，交易价值数亿美元，贾扬清及 Lepton 共同创始人白俊杰一起加入英伟达。 此前，贾扬清与白俊杰曾在 Meta、阿里云任职。 Lepton AI 成立于 2023 年，前有约 20 名员工，于 2023 年 5 月完成 1,100 万美元的天使轮融资，是贾扬清离开阿里后创办的 人工智能 基础设施 (AI Infra) 公司，总部位于美国加利福尼亚州帕洛阿托。 Lepton AI 的核心业务是围绕英伟达 GPU 服务器展开的一系列服务。一方面，Lepton AI 从事英伟达 GPU 服务器的出租业务。它并不直接拥有 GPU，也不自行管理数据中心或服务器，而是凭借独创的“云原生+多云整合”技术，从云提供商租用服务器后再转租给自己的客户，能够以极低成本调度全球 GPU 资源，满足客户在 AI 训练和推理方面的算力需求。 另一方面，Lepton AI 通过开发软件帮助初创企业在云中构建和管理自己的 AI 应用。其旗下的 Lepton AI Cloud 平台针对 AI 工作负载进行优化，用户可以使用它来训练 AI 模型和执行推理任务。该平台提供多种英伟达显卡供用户选择，还提供可视化界面用于设置训练集群。 对于需要训练多个模型的企业用户来说，这让他们能够集中管理分配给每个项目的硬件数量。在模型开发完成后，用户可将其部署在经过推理优化的实例上，Lepton AI 提供每秒超过 600 个 tokens 的处理速度，同时将延迟保持在 10 毫秒以下。 并且当需求增加时，Lepton AI 平台的扩展功能会自动为用户的推理环境添加更多硬件资源。 此外，Lepton AI 还使用 vLLM 等开源工具来加快推理速度。其客户包括月活数十万的 AI 游戏公司 Latitude.io 及科研平台 SciSpace 等。 总的来说，英伟达收购 Lepton AI 有助于整合双方在 AI 技术上的优势。Lepton AI 可能在特定 AI 算法、模型优化方面有独特之处，与英伟达强大的 GPU 计算技术相结合，可以进一步提升 AI 计算效率。 从市场角度看，这一收购能拓宽英伟达的市场覆盖。Lepton AI 或许在某些细分市场或行业应用中有自己的客户基础和渠道，英伟达借此可以深入这些领域，如新兴的 AI - 驱动的工业自动化或者特定的医疗影像分析市场，增强其在不同行业 AI 解决方案中的话语权。毫无疑问，此次收购也将为英伟达带来更多的人才资源。 贾扬清是华人领域比较知名的 AI 科学家。早年其毕业于清华，本科和硕士阶段均就读于信息科技自动化专业。研究生期间，他开始转向机器学习研究，这一选择纯粹出于兴趣——尽管当时机器学习并非热门方向，甚至面临“毕业即失业“的担忧。2009 年，他以《基于正则化的半监督学习研究》获得校级优秀硕士学位论文。 随后，他获得包括美国斯坦福在内的 5 所名校博士录取，最终在导师建议下选择美国加州大学伯克利分校计算机科学专业。 在伯克利攻读博士期间，贾扬清迎来了人生的第一个高光时刻。2013 年，他利用英伟达捐赠的一块 K20 GPU，在准备毕业论文的间隙开发出了深度学习框架 Caffe。这个起初只是“不想写论文“的业余项目，最终成为全球首个通用深度学习框架。 有趣的是，框架名称“Caffe“源于他在谷歌实习时喝太多咖啡的经历，取这个名字是为了提醒自己戒掉咖啡因。当他在导师“写大家都会用的系统比写无人关注的论文更重要“的鼓励下，其选择将 Caffe 完全开源，这一决定彻底改变了深度学习领域的研究生态。 Caffe 迅速被微软、雅虎、英伟达、Adobe 等公司采用，成为图像识别领域的标准工具 。2013 年博士毕业后，贾扬清加入谷歌大脑团队，成为 TensorFlow 的核心开发者之一 。2016 年，他转投 Facebook(现 Meta)，担任 AI 架构总监，领导开发了移动端深度学习框架 Caffe2Go，实现了 AI 模型在智能手机 CPU 上的实时运行。 在 Facebook 期间，他还推动了 Caffe2 框架的开源，使其成为 Facebook 机器学习的基础技术。这两段经历让他深入参与了全球最主流的两大深度学习框架(TensorFlow 和 PyTorch)的研发，奠定了他在 AI 系统架构领域的权威地位。 2019 年 ，贾扬清做出一个令业界瞩目的决定——加入阿里巴巴，担任技术副总裁，领导阿里云智能计算平台事业部。在阿里期间，他推动建设了大数据和 AI 平台，将相关业务做到行业领先位置。2022 年，他主导阿里达摩院与 CCF 开源发展委员会合作推出开源模型社区“魔搭”。 在阿里四年间，贾扬清实现了从 AI 框架专家到云计算业务领导者的角色转变。
Lepton AI Frequently Asked Questions (FAQ)
When was Lepton AI founded?
Lepton AI was founded in 2023.
Where is Lepton AI's headquarters?
Lepton AI's headquarters is located at 20863 Stevens Creek Boulevard, Cupertino.
What is Lepton AI's latest funding round?
Lepton AI's latest funding round is Acquired.
How much did Lepton AI raise?
Lepton AI raised a total of $11M.
Who are the investors of Lepton AI?
Investors of Lepton AI include NVIDIA, Charles River Ventures and Fusion Fund.
Who are Lepton AI's competitors?
Competitors of Lepton AI include OpenAI and 4 more.
Loading...
Compare Lepton AI to Competitors
evroc builds a hyperscale cloud for the digital future of Europe. Its offerings include cloud services that cover compute, storage, networking, and application integration, with attention to privacy and security. evroc serves sectors that need cloud infrastructure and services with data sovereignty and environmental responsibility. It was founded in 2022 and is based in Stockholm, Sweden.
Trellis Data specializes in artificial intelligence and machine learning solutions within the technology sector. The company offers a platform that provides explainable AI for real-world problem-solving in various industries, with a focus on speech-to-text, object detection, scene text reading, and mobile AI technologies. Trellis Data primarily serves sectors such as logistics, defense, law enforcement, and biosecurity. It was founded in 2017 and is based in Barton, Australia.
Anthropic operates a safety and research company focused on developing AI systems. The company's main offerings include Claude, an AI assistant for various tasks, and a suite of research initiatives aimed at AI safety and interpretability. Anthropic's research includes natural language processing, human feedback, reinforcement learning, and other areas. It was founded in 2021 and is based in San Francisco, California.
Inflection AI focuses on enterprise artificial intelligence, providing artificial intelligence (AI) model training and tuning while addressing data security and cost efficiency for enterprise clients. The company's solutions integrate into business operations. It was founded in 2022 and is based in Palo Alto, California.
Aleph Alpha specializes in generative artificial intelligence (AI) technology for enterprises and governments within the artificial intelligence sector. The company offers language models and multimodal AI solutions for human expertise, compliance, and AI applications. Aleph Alpha's technology is tailored for various architectures and value chains. It was founded in 2019 and is based in Heidelberg, Germany.
Sakana AI focuses on developing artificial intelligence (AI) through nature-inspired foundation models within the research and development sector. Its main offering includes creating a new kind of foundation model that draws inspiration from natural intelligence, designed to advance the field of AI. It was founded in 2023 and is based in Tokyo, Japan.
Loading...