Valued at $4 billion, Luma AI, invested by NVIDIA, announces a large-scale expansion plan, aiming at the "world model"!

In the fierce competition of the artificial intelligence sector, the Nvidia-backed startup Luma AI is expanding its frontlines from Silicon Valley in the US to Europe.

On Tuesday local time, the company, headquartered in Palo Alto, officially announced its large-scale expansion plan in London. According to the plan, Luma AI will recruit about 200 employees at its new London base, covering positions in research, engineering, partnerships, and strategic development, aiming for completion by early 2027. By then, the London team will account for approximately 40% of its global workforce.

Behind this strategic move is its strong capital backing. Just two weeks ago, Luma AI announced its completion of a $900 million Series C funding round led by AI company Humain under the Saudi Public Investment Fund, pushing its valuation past $4 billion. This funding provides ample ammunition for Luma AI's global expansion and computing infrastructure construction.

Luma AI’s expansion is not an isolated case; it joins a wave of North American AI companies flooding into the UK and Europe. This move is not only to compete for top talent but also to seize new market opportunities, and will directly intensify its head-to-head competition with tech giants such as Google and Meta in the domains of generative video and “world models.”

Huge Funding Supports Global Expansion

The recent $900 million funding completed by Luma AI is the key enabling its global expansion. This round was led by Humain, and the company had previously received backing from chip giant Nvidia. Luma AI’s CEO and co-founder Amit Jain stated:

“With our Series C funding and the imminent global computing infrastructure buildout, we now have the capital and capability to deliver world-class AI to creators worldwide.”

He emphasized that expanding into Europe and the Middle East is “the logical next step,” aiming to directly deliver AI technology to storytellers, agencies, and brands across the globe. Currently, the company’s video models are mainly provided to the marketing, advertising, media, and entertainment industries through application programming interfaces (API) and content creation suites.

Among the many European cities, Luma AI has chosen London as its first stop and core hub for expansion. In an interview, Amit Jain explained that this decision is based on the UK’s talent advantage. He said:

“Given the universities and research institutions like DeepMind here, London has some of the best research talent available.”

He also added: “We also see London as a gateway to the European market.”

Luma AI’s move reflects a clear industry trend. To capitalize on Europe’s talent pool and revenue opportunities, North American AI labs are all ‘setting up camp’ here. For example, San Francisco-based Anthropic announced plans last November to open offices in Paris and Munich; the Canadian AI startup Cohere said last September that it would establish its EMEA (Europe, Middle East, and Africa) headquarters in Paris.

Competing on the “World Model” Track

Luma AI’s core goal is to build “world models.” These models are capable of processing and learning from video, audio, image, and text information, and compared to large language models (LLM) that mainly handle text, their dimensional coverage is broader.

Although development in “world models” is not as advanced as with LLMs, some researchers believe they may be even more critical on the path to achieving Artificial General Intelligence (AGI).

Amit Jain predicts that although visual models currently lag behind language models by about one to one and a half years, over time, “world models” will become the “natural interaction interface” for most everyday AI use cases. He said that tech giants including Google, Meta, and Nvidia are developing “world models” for various use cases, making the competition in this domain extremely intense.

According to Amit Jain in media interviews, Luma AI's latest model Ray3, released in September, scored higher than OpenAI’s Sora in performance benchmarks and is at a similar level to Google’s Veo 3, demonstrating its determination to compete with industry giants in core technology.

Risk Warning and DisclaimerThe market contains risks; investment should be cautious. This article does not constitute personal investment advice, nor does it consider the individual investment objectives, financial circumstances, or needs of any specific user. Users should consider whether any opinions, views, or conclusions mentioned in the article are suitable for their particular situation. Investment based on this is at your own risk.