DZ
DefZone.Net

Kien thuc cong nghe & giai tri

Wiki Article

Draft:Anyscale

Nguồn dữ liệu từ Wikipedia, hiển thị bởi DefZone.Net

Xem trên WikipediaBài ngẫu nhiên khác
  • Comment: So what is needed here is truly independent and reliable sources talking about the company in sufficient depth to give "corporate depth". Within that some coverage can be deemed to be "corporate trivia", such as staff changes. See WP:SIRS for the definitions. Most of the sources here are not independent or reliable enough for that test, but in many cases they are ok for the separate issue of verification. ChrysGalley (talk) 07:54, 25 June 2026 (UTC)
  • Comment: In accordance with the Wikimedia Foundation's Terms of Use, I disclose that I have been paid by my employer for my contributions to this article. Ray Anyscale (talk) 07:45, 25 June 2026 (UTC)


Anyscale, Inc.
TypePrivate
IndustryArtificial intelligence, Distributed computing
Founded2019; 7 years ago (2019) in Berkeley, California[1]
FoundersRobert Nishihara
Philipp Moritz
Ion Stoica
Headquarters
San Francisco, California
,
United States
Websitewww.anyscale.com

Anyscale, Inc. is an American artificial intelligence infrastructure company headquartered in San Francisco[2], California. It develops an AI computing platform built around Ray, an open-source framework for distributed Python and AI workloads.[3] The company was founded in 2019 by Robert Nishihara, Philipp Moritz and Ion Stoica, the original creators of Ray at the University of California, Berkeley's RISELab.[1]

Anyscale incorporates the Ray distributed compute engine into a platform that includes Developer Central - suite of developer tools for development of AI workloads - and Cluster Controller - unified pane to manage, govern and monitor clusters across regions or clouds.[4] The company recently announced an agentic-forward development experience with Anyscale Agent Skills.[5]

The platform operates natively across clouds including Amazon Web Services, Google Cloud Platform, Microsoft Azure[6], CoreWeave [7], Nebius Group[8] & more.

Anyscale is used to run a wide range of AI workloads at Foundation model scale from multimodal data processing, to distributed training and large scale inference. This includes workloads such as embedding processing at scale at Notion [9], planet-scale image processing at Xoople [10], video processing and model training at Torc Robotics [11], distributed model training at Discord [12] and more [13].

History

[edit]

Anyscale traces its origins to the RISELab at UC Berkeley, where doctoral students Robert Nishihara and Philipp Moritz, working on solving the challenges of running reinforcement learning realized complexity of scaling infrastructure and began working on Ray project in 2016 to make it easier for Python developers to build distributed applications.[1] Ion Stoica, a computer-science professor who had previously co-founded Databricks, joined Nishihara and Moritz to incorporate Anyscale in 2019 with the goal of commercializing Ray and democratizing large scale AI.[1]

The company emerged from stealth in December 2019 with US$20.6 million in Series A funding led by Andreessen Horowitz, with participation from NEA, Intel Capital and Foundation Capital.[1] Since then Ray has grown into one of the most widely adopted open-source frameworks for distributed AI, used internally at large technology companies and AI laboratories.[14]

In July 2024 Anyscale announced that Keerti Melkote - the founder of Aruba Networks, which was acquired by Hewlett Packard Enterprise in 2015 - had been appointed chief executive officer.[15][16]

In November 2025, as Ray downloads grew past 7 million per week, Anyscale announced the donation of the Ray project to the PyTorch Foundation [17], home to some of the most widely adopted open-source AI projects including VLLM, DeepSpeed, and PyTorch. At that same time, Anyscale announced a strategic partnership and joint product engineering effort to make Anyscale an Azure native integration. [6]

Funding

[edit]

Anyscale launched in December 2019 with US$20.6 million in Series A funding led by Andreessen Horowitz.[1] In October 2020 the company raised an additional US$40 million Series B round led by NEA, bringing total funding at the time to roughly US$60 million.[18]

In December 2021 Anyscale announced a US$100 million Series C round at a US$1 billion valuation, co-led by Andreessen Horowitz and Addition, with participation from NEA, Intel Capital and Foundation Capital. The round brought total disclosed financing to approximately US$160 million.[19]

Ray Summit

[edit]

Anyscale also hosts Ray Summit, an annual industry conference in San Francisco that has featured speakers from Uber, OpenAI, Shopify, IBM and other organizations using Ray in production.[20]

In 2026, Ray Summit will be hosted in San Francisco at the Marriott Marquis hotel and will include speakers from Bedrock Robotics, Periodic Labs, Lila Sciences and more working on frontier AI including work on Physical AI, LLM reinforcement learning, multimodal data curation and foundation model training. [21]

References

[edit]
  1. ^ a b c d e f "Anyscale, from the creators of the Ray-distributed computing project, launches with $20.6M led by a16z". Anyscale. December 2019.
  2. ^ "About Anyscale". anyscale.com. Anyscale. Retrieved 2026-05-19.
  3. ^ "Scale Machine Learning & AI Computing". ray.io. Anyscale. Retrieved 2026-05-19.
  4. ^ "Anyscale Platform". anyscale.com. Anyscale. Retrieved 2026-06-24.
  5. ^ Geng, Kunling; Abiar, Aydin; Wang, Philip; et al. (April 22, 2026). "Introducing Anyscale Agent Skills: Build faster, debug smarter, and optimize AI workloads running on Ray". Anyscale Blog. Anyscale. Retrieved 2026-05-19.
  6. ^ a b "Anyscale on Azure". azure.microsoft.com. Microsoft. Retrieved 2026-06-24.
  7. ^ "Announcing Distributed AI on CoreWeave With Fully Managed Ray on Anyscale". coreweave.com. CoreWeave. Retrieved 2026-05-19.
  8. ^ "Nebius and Anyscale partner to power cost-efficient multimodal and physical AI". nebius.com. Nebius. Retrieved 2026-06-24.
  9. ^ "Two years of vector search at Notion: 10x scale, 1/10th cost". notion.com. Notion. Retrieved 2026-06-24.
  10. ^ "How Xoople Scales Python for AI using Anyscale on Azure". youtube.com. YouTube. Retrieved 2026-06-24.
  11. ^ "How Torc Robotics Scales Multimodal AI for Autonomous Driving with Ray". youtube.com. YouTube. Retrieved 2026-06-24.
  12. ^ "From Single-Node to Multi-GPU Clusters: How Discord Made Distributed Compute Easy for ML Engineers". discord.com. Discord. Retrieved 2026-06-24.
  13. ^ "Anyscale Customers". anyscale.com. Anyscale. Retrieved 2026-06-24.
  14. ^ "How Ray, a Distributed AI Framework, Helps Power ChatGPT". The New Stack. Retrieved 2026-05-19.
  15. ^ "Anyscale Names Industry Veteran Keerti Melkote Chief Executive Officer" (Press release). Anyscale. July 31, 2024. Retrieved 2026-05-19.
  16. ^ "Anyscale, software provider for OpenAI and other major firms, hires new CEO to boost sales". Semafor. July 31, 2024. Retrieved 2026-05-19.
  17. ^ "PyTorch Foundation Welcomes Ray to Deliver a Unified Open Source AI Compute Stack". pytorch.org. PyTorch. Retrieved 2026-06-24.
  18. ^ "Anyscale adds $40M to bring its Ray-based distributed computing tech to the enterprise masses". TechCrunch. October 21, 2020. Retrieved 2026-05-19.
  19. ^ "Anyscale Secures $100M Series C at $1B Valuation to Radically Simplify Scaling and Productionizing AI Applications" (Press release). BusinessWire. December 8, 2021. Retrieved 2026-05-19.
  20. ^ "Ray Summit 2024: Breaking Through the AI Complexity Wall". Anyscale. Retrieved 2026-05-19.
  21. ^ "Ray Summit 2026". anyscale.com. Anyscale. Retrieved 2026-06-24.