Center-left think tanks and other non-profit organizations have filed a petition with the European Commission over concerns that NVIDIA’s acquisition of workload management startup Run:ai could lead to increased control of the GPU giant over the AI industry. They are calling for a “thorough investigation.”
The purchase price, first revealed in April, has not yet been confirmed, but Tel Aviv-based AI-focused Kubernetes orchestration business Run.ai has raised $18 million in four funding rounds since its founding in 2016. It raised $700 million in cash and was valued at $700 million. 2018.
Run.ai’s platform provides a central Unser interface and control plane that allows customers to use different Kubernetes flavors, and incorporates some of the same tools for managing namespaces, resource allocation, and more. Masu.
As previously noted, a key feature of Run.ai is that the platform is built to integrate with third-party AI tools and frameworks for a GPU-accelerated container environment. It already supports Nvidia’s DGX computing platform, and Nvidia is committed to maintaining the same business model for Run.ai.
Nevertheless, the Open Market Institute and 18 other organizations sent a letter to the EC stating that the deal “reinforces NVIDIA’s super-advantage by integrating Run.ai software and providing GPUs. It called on competition authorities to investigate concerns that “the More barriers have been created around the chip empire. ”
France ready to bring ‘prosecution against NVIDIA’
read more
“We are a coalition of civil society organizations deeply concerned that this merger threatens to further exacerbate the extreme and growing concentration of power in the artificial intelligence technology stack, particularly in the provision of computing power.” This is stated in the submission (PDF).
In addition to the Open Markets Institute, other signatories include campaign groups Article19, FoxGlove, Balanced Economy Project, AI Now Institute, Gentium, and IT for Change.
By some (perhaps conservative) estimates, Nvidia already has a dominant market share in terms of supplying processors to train and deploy large-scale language models. Revenue from AI semiconductors is estimated to reach $71 billion in 2024, up 33% year over year.
Nvidia’s revenue for the nine months ended October 27, 2024 was $91.2 billion, compared to $38.82 billion in the year-ago period. Operating income increased from $19.35 billion to $57.4 billion.
The Open Market Institute’s submission to the EC calls on the EC to “launch a full investigation into the merger, focusing on the risks it poses to the AI sector as a whole”.
“The merger will strengthen NVIDIA’s control over the supply of advanced chips, which are critical inputs to downstream markets, particularly cloud and AI,” it added. “Despite its efforts to develop its own competing chips, NVIDIA supplies both leading cloud providers as well as smaller “neo-cloud” startups such as Coreweave, Crusoe, and Lambda1. This makes it a key input provider across the AI sector. All of them then provide cloud computing and AI services to a wide range of companies in Europe and around the world. ”
Nvidia is “fully booked” in terms of inventory sales allocations for next year, and as we’ve already noted, is the kingmaker in terms of which companies will or will not get their chips.
The submission continues as follows.
According to the letter, the signatories say they estimate that Nvidia has an 88% market share of GPU sales worldwide, leaving Intel and AMD to suffer. Nvidia is using CUDA to keep competitors out, and is “developing software products to build a moat around its chips, creating a closed environment with tightly integrated GPUs and proprietary software.” “I do,” he claims.
The letter continues: “This is why Nvidia is acquiring Run:ai. By optimizing the use of scarce computing resources, Run:ai’s GPU orchestration and virtualization services will Orchestration using GPUs can significantly improve the performance of It is already considered a key asset by GPU users because it reduces the computational costs associated with development and scales resources across workloads to maximize performance and reduce costs. It’s standard market practice. ”
The letter states that barring access to Run.ai from rivals is a major concern, and that Nvidia has “the ability and incentive to tie or bundle Run:ai’s services with its own GPUs.” It is stated that.
The letter adds, “The Committee should not be misled by Nvidia’s announcement that the Run:ai service will remain open source.”
“There are varying degrees of openness, and AI companies often use ‘open’ rhetoric in ways that actually strengthen their market power. For example, ‘open source’ can include the use of competing products and Such an approach would prevent Nvidia from restricting access to the relevant technology or cutting it off entirely at a later stage. Not. ”
In a statement to The Register, an Nvidia spokesperson said:
“We look forward to welcoming Run:ai employees to our team.Run:ai currently supports only NVIDIA GPUs, but once the regulatory process is complete, we will open source Run:ai and provide all Not surprisingly, we haven’t heard any complaints from customers, partners, or even competitors about our plans. ”
The European Commission told us that the deadline to “make a decision” on NVIDIA’s acquisition of Run.ai is December 20th, and that it would not be commenting further at this stage. ®
