AMD is acquiring World Labs, the spatial AI company co-founded by Stanford researcher Fei-Fei Li, in an all-stock deal valued at $8.2 billion. Li will join AMD as executive vice president and chief scientist.

The deal is AMD’s largest acquisition since it bought Xilinx for $49 billion in 2022, and it signals the company’s bet that spatial AI, the ability to generate and understand 3D environments, will be critical to everything from robotics to gaming to autonomous systems.

World Labs launched in 2024 and hit a $1 billion valuation within months. It shipped its first commercial product, Marble, in 2025. Marble lets users create interactive 3D worlds from text prompts, a capability that has applications in game development, architectural visualization, and simulation environments for training AI systems.

Li is one of the most cited researchers in computer vision. She led the ImageNet project, which became the foundation for the deep learning revolution of the 2010s. Her move to AMD brings credibility to a company that’s been fighting to close the gap with Nvidia in AI hardware.

AMD’s GPU business has grown, but it still lags far behind Nvidia in the data center AI market. World Labs gives AMD a differentiated software layer and a high-profile research leader who can attract talent and customers.

The timing is notable. Nvidia just launched a new platform for securing AI agents, a toolkit designed to keep agents from breaking out of their test environments. Jensen Huang introduced the system on Monday, responding to a series of safety incidents involving rogue agents. AMD now has Li, whose work on responsible AI and computer vision gives the company a credible voice in the safety conversation.

Meta finally goes after enterprise

On the same day, Meta announced it’s launching an enterprise AI platform and hired MongoDB CEO Dev Ittycheria to lead the effort.

This is Meta’s most serious attempt yet to sell software directly to businesses. The company has built consumer AI products, released open-source models, and made its Llama family of models available for commercial use. But it hasn’t operated a true enterprise sales organization.

That changes now. The new platform brings together Meta’s full stack: Muse (its all-in-one AI agent), Meta Business Agent, Muse API, and Muse Code. Ittycheria, who spent more than a decade building MongoDB into a multi-billion-dollar enterprise company, will oversee the business.

Meta’s advantage is its model quality and infrastructure. Llama 4 is competitive with closed models from OpenAI and Anthropic, and Meta has the scale to offer aggressive pricing. The company has also been integrating AI into its own products, WhatsApp and Instagram, which gives it real-world data on what works.

The disadvantage is trust. Enterprises have been reluctant to adopt Meta’s tools because of the company’s privacy record and its focus on advertising. Hiring Ittycheria is a signal that Meta understands this problem and is willing to invest in solving it.

Other deals and moves

Instinct, the viral AI agent, raised a $1 billion Series C at a $10 billion valuation. Founder Noah Shinn said the funding will help the company scale. Instinct became one of the breakout consumer AI products of 2025, offering a persistent agent that manages tasks across apps and devices.

Modal Labs, an inference provider, is closing a $750 million round at a $15.75 billion valuation, according to a source. That’s more than triple its valuation from four months ago, reflecting the explosion in demand for inference infrastructure.

Peak XV, the venture firm formerly known as Sequoia India, raised the ceiling for its Surge seed program to $5 million and unveiled an 18-startup cohort. Thirteen of the companies are targeting global markets. More than half are based in India.

Outmarket, an insurtech startup that uses AI to automate paperwork for insurance brokers, raised $34.5 million just months after its prior round.

And Anthropic filed its prospectus for a potential IPO, revealing that it’s losing tens of billions of dollars a year but growing rapidly. The company also included a warning that its AI systems could pose an existential risk to humanity, a disclosure that’s likely to get scrutiny from regulators and investors.

The AMD and Meta moves are the bigger story. AMD is betting billions that spatial AI will matter. Meta is betting that it can convince enterprises to trust it. Both are expensive bets, and both will take years to pay off.