[HN Gopher] Fei-Fei Li's World Labs raised $1B from A16Z, Nvidia...
___________________________________________________________________
Fei-Fei Li's World Labs raised $1B from A16Z, Nvidia to advance its
world models
Author : aanet
Score : 57 points
Date : 2026-02-18 17:18 UTC (5 hours ago)
(HTM) web link (www.bloomberg.com)
(TXT) w3m dump (www.bloomberg.com)
| darth_avocado wrote:
| Headline suggests A16Z, Nvidia as investors, but the article and
| their own blog suggests Autodesk, AMD etc. as the bigger
| investors
| giwook wrote:
| Clickbait.
| iwontberude wrote:
| This site probably has fans or bots of the egg head submitting
| and upvoting things.
| giwook wrote:
| https://archive.is/qjsFy
| chaosprint wrote:
| related:
|
| https://news.ycombinator.com/item?id=45923326
|
| https://entropytown.com/articles/2025-11-13-world-model-lecu...
|
| I still want to know the definition of a world model, because it
| seems that what they are doing now is a very basic model of GS
| (not sure if it is self-developed) and a GS browser.
| tintor wrote:
| World model is a 3d game engine, which uses neural net to
| compute screen pixels instead of rasterizing triangles with
| textures or ray-tracing.
|
| It is different from video generators, because it also accepts
| control inputs (ie. keyboard, mouse).
| janalsncm wrote:
| I think a robot would want to know a little more about the
| world though? For instance, picking up a spatula requires
| less force than picking up a an anvil. So you would want to
| know about the mass of things.
| Aboutplants wrote:
| What are the industries that would truly benefit from good world
| models? Obviously there are benefits for Robotics and Self
| Driving but what else am I missing?
| KaiserPro wrote:
| depending on what you are using them for they are quite good
| for planning 3d operations.
|
| However the definition of "world models" is a bit nebulous.
| World labs looks kinda like sprites in a 3d world.
|
| Which also leads me to suggest that in no way are they worth
| $1b, espcially as google are doing to destroy them (possibly
| meta, but I doubt they are going to be any good anytime soon)
| jeffjobs4000 wrote:
| Robotics is an enormous TAM, and would be all this company
| needs to justify this type of valuation.
|
| Also, video games, and movie / tv / video content generation.
| adityashankar wrote:
| Robotics benefits nearly every industry dependent on robotics,
| being able to expand your market base to countries where
| individuals have lower incomes obviously provides a boost to
| companies that wish to sell more product.
|
| It also allows for companies to reclaim their supply chains
| within the country of manufacture (for high income countries)
| aspenmartin wrote:
| World models are not a new idea, it comes from the "model-free"
| and "model-based" reinforcement learning paradigms that have
| been around forever
|
| Model-free paradigms (which we do now without world models)
| does not actually model what _happens_ when you take an action,
| they simply model how good or bad an action is. This is highly
| data inefficient but asymptotically performs much better than
| model-based RL because you don't have modeling biases.
|
| Model-based RL, where world-models come in, models the
| transition matrix T(s, a, s') meaning, I'm in state s and I
| take action a, what is my belief about my new state? By doing
| this you can do long-term planning, so it's not just useful for
| robotics and video generation but for reasoning and planning
| more broadly. It's also highly data efficient, and right now,
| for robotics, that is absolutely the name of the game.
|
| What you will see is: approximately zero robots, then
| approximately one crappy robot (once you get performance +
| reliability to jusssst cross the boundary where you can market
| it, even at a loss! and people will buy it and put it in their
| homes). Once that happens you get the magic: data flywheel for
| robotics, and things start _rapidly_ improving.
|
| Robotics is where it is because it lacks the volume of data we
| have on the internet. for robotics today it's not only e.g.
| egocentric video that we need but also _sensor-specific_ and
| _robot-specific_ data (e.g. robot A has a different build +
| components than robot B)
| alyxya wrote:
| The interesting thing to me about their world models is that it's
| like a static point cloud model, as opposed to deepmind's genie 3
| which feels more like a video generation model at its base. I see
| the video generation base as generally superior but far more
| expensive. The current approach for world labs is likely based on
| the expertise of the founders, but I don't see how it can scale
| and match what genie 3 does even if it does offer state
| persistence as an advantage.
| isusmelj wrote:
| Yes, Marble (from World Labs) feels like it's generating
| Gaussian Splats or similar. I guess it's more compatible and
| easier to use for 3d asset generation and reusing in other
| software. Very exciting times ahead!
| agentifysh wrote:
| I am not trying to be mean but this does not smell right to me,
| getting a solution too early for a problem vibes.
| seg_lol wrote:
| NVidia has to spend the money while the music still plays.
| NVidias money doesn't exist. Nothing has meaning, the value of
| money depends on who holds it.
___________________________________________________________________
(page generated 2026-02-18 23:01 UTC)