Skip to content

Nvidia in talks to buy Reflection AI or invest more, after putting in $800 million

The Financial Times says Nvidia is weighing a bigger stake in, or a takeover of, the open-model start-up founded by two former DeepMind researchers. An "acqui-hire" is one option.

Published 2 min read
WhatsApp
Nvidia headquarters in Santa Clara with its white steel lattice canopy and Nvidia sign
Nvidia headquarters in Santa Clara with its white steel lattice canopy and Nvidia signPhoto: Coolcaesar / Wikimedia Commons, CC BY-SA 4.0

Nvidia is in talks to put more money into Reflection AI or buy the company outright, the Financial Times reported on Saturday, citing people with direct knowledge of the discussions. Reuters' summary, carried by Business Recorder, says the talks are at an early stage and could still fall apart, though a deal could come within weeks.

Nvidia has already invested $800 million in the start-up, according to the FT. Reflection declined to comment and Nvidia did not respond straight away.

What kind of deal

The FT said several structures are on the table, including an "acqui-hire": Nvidia would hire Reflection's staff and license its technology without buying the company itself. Big tech firms have used this route repeatedly since 2024 because it can avoid a long competition review, which a full takeover by the world's most valuable chipmaker would almost certainly face.

Who Reflection AI are

  • Founded: 2024, by former Google DeepMind researchers Misha Laskin (chief executive) and Ioannis Antonoglou.
  • What it builds: tools that automate software development, one of the fastest-growing uses of AI.
  • Its model: on Monday it released Beam, its first open-weight model, aimed at coding and agent tasks. Open-weight means anyone can download and run it.
  • Valuation: Laskin told CNBC in April the company was raising money at a $25 billion pre-money valuation; it is not known whether that round closed.

Reflection pitches itself as an American answer to cheaper Chinese open models such as DeepSeek and Kimi.

Why Nvidia wants it

Nvidia sells the chips that train and run almost every large AI model. Backing open-weight models built in the US gives companies and governments a reason to keep buying its hardware rather than relying on closed models from a handful of labs, or on Chinese alternatives that run on other chips.

What it means for developers in Pakistan

Open-weight coding models matter most where paid AI subscriptions are expensive in rupees. A model like Beam can be run on a local server or a rented GPU and paid for once, rather than per seat in dollars. Freelancers billing foreign clients can check what they owe on that income with our freelancer tax calculator, and convert dollar invoices with the currency converter at today's rate of Rs 277.00 to the dollar.

For more on where AI is heading this week, see Satya Nadella's call for an AI emergency brake.

Frequently asked questions

How much has Nvidia invested in Reflection AI?
Nvidia has already invested $800 million in Reflection AI, according to the Financial Times report of 10 October 2026.
What does Reflection AI make?
Reflection AI, founded in 2024 by former DeepMind researchers Misha Laskin and Ioannis Antonoglou, builds tools that automate software development and released its first open-weight model, Beam, for coding and agent tasks.
Comments
WhatsApp

Sources

Tags:artificial intelligence

More in Technology

Reader comments 0

No comments yet. Say something useful.