344
Productivity & Workflow355
Automation & Workflow224
Software Development251
Marketing & Growth192
AI Infrastructure & MLOps174
Writing & Content Creation203
Data & Analytics141
Design & Creative170
Photography & Imaging156
Customer Support131
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
Reflection AI agreed to a $1 billion deal with Nebius to access Nvidia chips and computing capacity needed to build and run its open AI models.
In short: Reflection AI has signed a $1 billion deal with Nebius to get the computing power it needs to build and run its AI models.
Reflection AI, a US startup founded in 2024, signed a $1 billion “compute” deal with Nebius, a European company that sells AI infrastructure. “Compute” is basically rented computer power (like renting a lot of very fast ovens to bake more bread at the same time).
Nebius will give Reflection access to Nvidia’s latest chips, which are specialized parts used to train and run AI systems. Training an AI model means running huge amounts of calculations until the system learns patterns from data, and that requires many chips working for long periods.
The Nebius agreement comes just weeks after Reflection signed another compute deal, this time with SpaceX, to use its computing resources. These kinds of partnerships have become common as AI companies compete to lock in enough capacity to keep improving their models.
Reflection is working on “open” AI models, meaning the model details, and sometimes the weights (the stored numbers that make the model work), can be shared publicly. The company is valued at about $8 billion, and it has raised close to $2.6 billion from backers including Nvidia, Sequoia Capital, and Lightspeed Venture Partners.
For everyday users, these deals can shape which AI tools are available and how reliable they are. If a company cannot get enough computing power, its products can be slower to improve, more expensive, or limited in who can use them. Interest in open models is also rising because some people and organizations worry that access to closed AI systems could be restricted suddenly due to business or government decisions.
Source: TechCrunch AI