5 Sources
[1]
Nscale buys Anyscale as it seeks to own more of the AI compute stack
In a bid to capture more of its customers' AI spending, British AI neocloud Nscale is buying software startup Anyscale, which helps companies scale their AI workloads across data centers and servers. Nscale is paying $1.65 billion for Anyscale, Bloomberg reported, citing an anonymous source. Founded by the same team that built the open-source Project Ray distributed programming Python framework, Anyscale started by building a platform that allowed people to run projects that needed large amounts of computing power. But after the launch of GPT-3 in 2022 brought AI into the spotlight, the company pivoted to offer scaling services for serving and training large language models, data curation, inferencing, reinforcement learning, and other tasks. The company's platform is built around Ray, offering developer tools, observability and orchestration. A deal to buy Anyscale would fall neatly into Nscale's focus on building vertically to serve compute needs. The neocloud has set up business lines across energy, data centers, orchestration software, and now, with Anyscale, it will also offer workload management and scaling. "Together, Anyscale and Nscale can co-design the software layer and infrastructure beneath it, something that neither company could do as effectively by optimizing its layer alone," Anyscale said in a statement. Nscale this March raised $2 billion in a Series C round that saw it valued at $14.6 billion. Its investors include Nvidia, Nokia, Blue Owl, Dell, and Norwegian industrial giant Aker. The neocloud has been busy putting that money, as well as various debt raises, to work, securing compute and data center partnerships with the likes of Microsoft, British Telecom, and Nordcraft. Anyscale, which was valued at $1.38 billion in a 2022 Series C round, said its revenue increased by 70% in its most recent quarter, compared to the previous sequential quarter. Nscale said Anyscale will continue to operate under its own branding and serve its existing customers. The startup's about 200 employees are all joining Nscale.
[2]
AI cloud provider Nscale to buy software startup Anyscale
July 30 (Reuters) - Cloud infrastructure provider Nscale will buy AI software startup Anyscale to help clients better manage and run AI systems efficiently, the companies said on Thursday. Bloomberg News first reported the news, saying the deal price is about $1.65 billion, citing a person familiar with the matter. Nscale declined to disclose the financial terms. Here are some details: Reporting by Jaspreet Singh in Bengaluru; Editing by Pooja Desai Our Standards: The Thomson Reuters Trust Principles., opens new tab
[3]
Nscale buys Anyscale for about $1.65bn to move up the AI stack
Nscale, the London AI-cloud company that generates its own power, has agreed to buy Anyscale, the startup behind the widely used open-source Ray software. Bloomberg put the price at about $1.65bn, a figure the companies did not disclose. The deal moves Nscale up from renting raw GPUs into the software layer that decides how much work each chip does, part of a land grab among the "neoclouds." Nscale, the London AI-cloud company that owns its own power and data centres, is buying its way up the stack. It has agreed to acquire Anyscale, the startup behind the widely used Ray software. Bloomberg reported the deal at about $1.65bn. Nscale announced the acquisition on Thursday but did not put a price on it. Bloomberg gave the figure, citing a person familiar with the deal. Anyscale's 200 or so staff, spread across the US, Europe and India, will join Nscale. The deal should close in the second half of 2026. Why a power company buys software Nscale's pitch has always been that it owns the layers beneath the chip: the electricity, the buildings, the racks. Buying Anyscale moves it up into the software that decides how much useful work each expensive GPU actually does. The logic is about margins. Renting raw GPUs by the hour is a commodity business, fought against every rival buying the same Nvidia systems. The money sits a layer up, in how a job gets split across thousands of chips at once. Anyscale's platform, built on the open-source Ray framework, does exactly that. It spreads data preparation, training, fine-tuning and inference across a whole fleet. The company claims that cuts total cost of ownership by up to 90% against a patchwork of separate tools. "Most infrastructure providers just buy GPUs and rent them," said Josh Payne, Nscale's chief executive and founder. He said Nscale instead builds and owns every layer itself: the power, the data centres, the compute and the software. Its product chief, Dan Bathurst, put the pitch plainly. An engineer can now come to Nscale, he said, as "a one-stop shop for all of their training, fine-tuning and inference services." The neocloud land grab Nscale is one of the "neoclouds," the wave of AI-focused data-centre newcomers chasing the boom. It grew out of a cryptocurrency-mining business in early 2024. It has moved fast since, passing 1GW of capacity and signing big compute deals. It has also lined up the money to keep building. In July it closed a $900m revolving credit facility, on top of a $2.5bn commitment to UK data centres. It plans to float, possibly later this year. Its board includes the former Meta executives Sheryl Sandberg and Nick Clegg. Its rivals are making the same move. In May, Nebius spent $643m to buy Eigen AI, another firm that makes AI run more cheaply. The chipmaker Qualcomm bought the compiler startup Modular this month. "Everyone's wanting to move up the stack," Bathurst told Bloomberg. The prize is to be the one-stop shop as the money floods into AI infrastructure. What Nscale is actually buying The open-source project at the centre of the deal is not for sale. The PyTorch Foundation took over Ray in 2025, and it had already logged some 237 million downloads. It stays community-governed and free to run anywhere, from Cursor to xAI. Nscale is buying the commercial platform, the engineering team and the customers, which include Coinbase, Runway and Bedrock Robotics. It will join the foundation too. Anyscale arrives with momentum, reporting 70% revenue growth in its most recent quarter. Its chief executive, Keerti Melkote, called the combined company "the first full-stack AI hyperscaler." That is the bet: own everything from the electricity to the trained model. It is the same logic behind the race to build national AI factories. Whether it pays off is a question of timing. Anyscale sells against capacity that runs today. Nscale's largest deployment, a 1.35GW site for Microsoft in West Virginia, only starts arriving in 2027.
[4]
Nscale buys AI infrastructure optimization startup Anyscale for reported $1.65B
Data center builder Nscale Global Holdings Ltd. today announced plans to acquire Anyscale Inc., a venture-backed provider of artificial intelligence software. The terms of the deal were not disclosed. Bloomberg cited a source as saying that Nscale is paying $1.65 billion. In March, the company raised $2 billion from a consortium that included Nvidia Corp. and other major tech firms. London-based Nscale is building a network of AI data centers that will span a half-dozen countries. Its flagship development is a 2,250-acre campus in Mason County, West Virginia that features its own miniature power grid. The company estimates that the site can theoretically host more than eight gigawatts worth of computing capacity. Anyscale, the startup that Nscale has agreed to buy, was launched in 2019 by a group of computer scientists. It commercializes an open-source project called Ray that was created by its founding team. The software is used to optimize large-scale AI clusters. Distributing an AI workload across multiple servers is not as simple as splitting it into smaller tasks and running each one on a separate machine. Developers must make numerous optimizations to ensure that the cluster will run reliably. Additionally, they must fine-tune its performance to avoid cost overruns and underutilized hardware. Ray eases the task. One of the chores that the software streamlines is mitigating hardware errors. If one of the servers in a cluster fails during an AI training run, Ray can automatically replace it with a new machine. That removes the need to write custom fault tolerance workflows. The software also helps developers reduce bandwidth costs. If an AI model and the dataset that it uses are stored on two separate servers, the machines must regularly exchange information over the network. Ray installs AI models and their datasets on the same machine to avoid unnecessary traffic. Anyscale commercializes the project with a paid cloud-based version. The service provides access to managed Ray clusters that developers can launch in under a minute. Additionally, it features monitoring dashboards that ease the task of troubleshooting technical issues. Nscale expects to close the acquisition by year's end. The company plans to offer Anyscale's software to its customers alongside its internally developed infrastructure optimization tools. Nscale provides managed versions of Kubernetes and Slurm, two open-source tools likewise designed to automate server cluster management. "We build and own every layer ourselves: the power, the data centers, the compute, and the software that turns them into an AI cloud," said Nscale founder and Chief Executive Officer Josh Payne. "Anyscale extends that offering with managed services that AI teams use to scale any workload, completing a truly vertically integrated AI cloud platform."
[5]
AI cloud provider Nscale to buy software startup Anyscale
July 30 (Reuters) - Cloud infrastructure provider Nscale will buy AI software startup Anyscale to help clients better manage and run AI systems efficiently, the companies said on Thursday. Bloomberg News first reported the news, saying the deal price is about $1.65 billion, citing a person familiar with the matter. Nscale declined to disclose the financial terms. Here are some details: o San Francisco-based Anyscale provides software to organize and manage complex AI tasks so they can run smoothly across many computers simultaneously. o Nscale said it expects to attract more customers by offering Anyscale's software. o Anyscale will continue to operate under its brand and serve its existing customers, the companies said. o Anyscale's entire team of about 200 people across the United States, Europe and India will join Nscale as part of the acquisition. o The deal is expected to close in the second half of this year, the companies said. o Founded in 2024, Nscale is known as a "neocloud," a vertically integrated AI cloud platform that owns and operates its own data centers, GPUs and software stack to deliver large-scale, GPU-powered AI compute. o Nscale competes with companies such as CoreWeave and Nebius Group. o Anyscale generated 70% revenue growth sequentially in the company's most recent quarter, according to a blog post by its co-founders. (Reporting by Jaspreet Singh in Bengaluru; Editing by Pooja Desai)
Share
Copy Link
British AI cloud provider Nscale is buying AI software startup Anyscale for $1.65 billion to expand beyond GPU rentals into workload management. The deal combines Nscale's ownership of power, data centers, and compute with Anyscale's Ray-based platform that optimizes AI workloads across thousands of chips, creating what the companies call the first full-stack AI hyperscaler.
British AI cloud provider Nscale is acquiring AI software startup Anyscale for approximately $1.65 billion
1
2
, marking a strategic shift from renting raw computing power to controlling the entire AI compute stack. The deal, expected to close in the second half of 2026, brings Anyscale's 200-person team across the United States, Europe, and India into Nscale's operations5
. While Nscale declined to disclose financial terms officially, Bloomberg reported the $1.65 billion price tag citing a person familiar with the matter3
.
Source: TechCrunch
The acquisition represents a land grab among neoclouds competing to capture more customer spending by owning multiple layers of AI infrastructure. Nscale, which raised $2 billion in a Series C round in March at a $14.6 billion valuation, has already secured power, data centers, and GPU capacity
1
. Adding Anyscale's workload management platform completes what CEO Josh Payne describes as building and owning every layer: the power, the data centers, the compute, and the software4
.Anyscale was founded in 2019 by the team that created the open-source Project Ray distributed programming Python framework
1
. The Ray software, which has logged 237 million downloads and moved to the PyTorch Foundation's governance in 2025, helps companies distribute AI workloads efficiently across thousands of servers3
. The platform handles data preparation, AI model training, fine-tuning, and inference services while reducing infrastructure complexity.
Source: Reuters
The software tackles critical AI infrastructure optimization challenges that make or break cost efficiency. Ray automatically replaces failed servers during training runs, eliminating the need for custom fault tolerance workflows
4
. It also reduces bandwidth costs by installing AI models and their datasets on the same machine to avoid unnecessary network traffic. Anyscale claims its platform cuts total cost of ownership by up to 90% compared to using separate tools3
.After GPT-3's launch in 2022 brought AI into the spotlight, Anyscale pivoted to offer scaling services specifically for large language models, reinforcement learning, and other AI tasks
1
. The company's commercial cloud-based version provides access to managed Ray clusters that developers can launch in under a minute, alongside monitoring dashboards for troubleshooting4
.Nscale's acquisition strategy reflects a broader trend among AI cloud providers racing to control more of the stack. The company positions itself as different from competitors who simply buy GPUs and rent them out. "We build and own every layer ourselves: the power, the data centers, the compute, and the software that turns them into an AI cloud," Payne explained
4
. This vertical integration approach aims to capture margins that commodity GPU rentals cannot deliver.
Source: SiliconANGLE
The London-based neocloud has moved aggressively since growing out of a cryptocurrency-mining business in early 2024
3
. It has surpassed 1GW of capacity and secured major compute partnerships with Microsoft, British Telecom, and Nordcraft1
. Its flagship development is a 2,250-acre campus in Mason County, West Virginia, featuring its own miniature power grid capable of hosting over eight gigawatts of computing capacity4
. A 1.35GW site for Microsoft in West Virginia is scheduled to come online in 20273
.Nscale's investors include Nvidia, Nokia, Blue Owl, Dell, and Norwegian industrial giant Aker
1
. The company closed a $900 million revolving credit facility in July, on top of a $2.5 billion commitment to UK data centers, and plans to float publicly possibly later this year3
. Its board includes former Meta executives Sheryl Sandberg and Nick Clegg.Related Stories
The combination creates what Anyscale CEO Keerti Melkote calls "the first full-stack AI hyperscaler"
3
. Engineers can now access a one-stop shop for training, fine-tuning and inference services, according to Nscale product chief Dan Bathurst3
. The deal positions Nscale to compete more effectively against rivals like CoreWeave and Nebius Group5
.Anyscale brings proven commercial traction, with customers including Coinbase, Runway, and Bedrock Robotics
3
. The company reported 70% revenue growth in its most recent quarter compared to the previous sequential quarter1
. Anyscale was valued at $1.38 billion in a 2022 Series C round, making the $1.65 billion acquisition price a modest premium that reflects both market conditions and strategic value1
.The open-source Ray project remains community-governed and free to run anywhere, from Cursor to xAI
3
. Nscale is acquiring the commercial platform, engineering talent, and customer relationships while joining the PyTorch Foundation. Anyscale will continue operating under its own brand and serving existing customers5
. Nscale plans to offer Anyscale's software alongside its internally developed infrastructure optimization tools, including managed versions of Kubernetes and Slurm4
."Together, Anyscale and Nscale can co-design the software layer and infrastructure beneath it, something that neither company could do as effectively by optimizing its layer alone," Anyscale stated
1
. This co-design capability represents the core bet: that controlling everything from electricity generation to trained models delivers better economics than competing at any single layer. Rival neoclouds are making similar moves, with Nebius spending $643 million to buy Eigen AI in May and Qualcomm acquiring compiler startup Modular this month3
. Watch whether this vertical integration strategy delivers the promised cost advantages as capacity comes online over the next year.Summarized by
Navi
[3]
[5]
09 Dec 2024•Business and Economy

09 Mar 2026•Startups

09 Jan 2026•Business and Economy

1
Technology

2
Technology

3
Technology
