Congratulations to @224ventureshq co-founder @ylecun on being named to the @TIME 100 AI 2026 list. Yann has spent four decades quietly building the foundations the rest of the field now stands on — from convolutional networks to his current bet, at Advanced Machine Intelligence, that world models are the path beyond today's LLMs. What TIME captures, and what I get to see up close, is his conviction that the next chapter of AI won't be written by scaling what already works. It'll come from researchers willing to question the paradigm and build something new. Lucky to be building alongside him. More to come from 224.

It was great to be back at @bloomberg studios to chat with @TimStenove1g. We covered quite a bit, but there's one thread worth amplifying for founders: the growing importance of having a non-consensus view. Essentially, an idea that doesn't seem plausible at first. People disagree with you. Investors leave the conversation disagreeing with the premise. But as the discussion progresses, the hypotheses that need to be tested become clear — and the result can be a category-defining company. When ChatGPT launched in November 2022, startups raced to embrace the technology and build applications that weren't possible before. Fairly obvious applications, like note-taking, were pursued by many. But we're almost four years into this inflection point — the same ideas have been pursued by dozens (hundreds?) of teams — and yet 99% of entrepreneurs still have consensus ideas. To be clear, you can pursue note-taking. Companies like @WisprFlow recently launched a note-taking app with clear reasons to win. But a consensus idea needs some special sauce — a unique angle that will open up, or create, a new category for your company. We're always connecting with founders to discuss their ideas. Send me a dm.

Here’s a piece by @goodfellow_ian, @sunfanyun, and me arguing that use of symbolic representations and game virtual world data offers the best path to building action-conditioned multimodal world models that enable reliable prediction and planning for long-horizon tasks. The advantages of symbolic language representations in this context was also recently argued for in new work from @physical_int: .

Introducing a world built by the Moonlake's world model. 🏙️ Most world models only allow for a limited action space. Moonlake maintains multimodal states across physics, appearance, geometry, and casual effects and predict how they evolve under different actions. 👇

We have big news to share: marimo is joining @CoreWeave! We're doubling down on open-source and scaling molab with serious compute Our mission is the same: to build the world’s best open-source notebook for working with data Read the full announcement:

Excited to see @sunfanyun & @sharonal_lee launching @moonlake! I got to know them at @Stanford, both doing super-exciting work with 3D world foundation models. Games and simulations need not just visuals but world understanding, and that's where they're focused. Take a look!

I told my parents that I’d like to drop out of my cs phd program at Stanford a few months back. They didn’t let me, we’re asian :) So I graduated and started @moonlake. @sharonal_lee and I saw urgency, and opportunities. Excited to share that we raised a 28 million dollar seed round to build the future for simulations and games. Grateful for the angels @naval, @goodfellow_ian @stevechen, @JeffDean @rauchg, @emerywells, @JaredLeto, @chrlaf, alongside many more, and the venture partners that we are fortunate to work with: @moislamvc, Shaun Johnson, @chrmanning, Artem Barsukov, Elvin Hao, @mercebent, @veelarco and William Freiberg. If you're a founder and you're not partnering with them, you're making a big mistake. Check out what we're about 👇

We raised $28M seed from Threshold Ventures, AIX Ventures, and NVentures (Nvidia's venture capital arm) —alongside 10+ unicorn founders and top AI researchers— to build reasoning models that generate real-time simulations and games. Models are bottlenecked by practical simulations that can act as Reinforcement Learning environments. Human self-expression is bounded by tools that let us create alternate realities. At Moonlake, we are building a future where anyone can create interactive worlds, bring their child-like wonder to life, learn within them, and most importantly, share experiences with people we care about. More in 🧵

Thrilled to see former @StanfordAILab students @sunfanyun and my own student @sharonal_lee today launching @moonlake, a new AI lab that lets anyone code interactive worlds in minutes. @aixventureshq led their $28M seed along @thresholdvc, @NVIDIA Ventures, and leading angels including @stevechen and @JeffDean.  @RashiShrivast18 at @Forbes covers the news: