AMD acquires World Labs for $8.2B San Francisco AI startup World Labs, founded by Fei-Fei Li (the “godmother of AI”), builds spatial intelligence models that perceive, generate, and interact with 3D virtual and physical worlds. Flagship product Marble turns text, images, or video into navigable 3D environments. New model Atlas adds camera-controlled generation and real-to-sim capabilities. AMD is buying the company in an all-stock deal worth ~$8.2 billion, expected to close by end of 2026. Fei-Fei Li will join as EVP & Chief Scientist. Spatial intelligence just leveled up. 🚀 #WorldLabs# #AMD# #AI# #SpatialIntelligence#

High-resolution #dToF# #LiDAR# is bringing more spatial intelligence to the edge enabling more accurate depth perception, improved object detection, and real time decisioning for robotics, smart vehicles, and industrial automation. Discover how🔍

Our collaboration with MaprGo unlocks structured spatial intelligence that powers Physical AI applications. By generating OpenSCENARIO datasets, building simulation-ready environments, or creating data components used by Physical AI systems, NATIX multi-cam data is the key. 🔑

TL;DR Treating spatial reasoning as a classification problem over finite candidates, instead of generating text, speeds up a VLM by up to 5.6x while also improving accuracy. Title: MJev-Spatial: System One Spatial Intelligence via Native Attention Matching URL: Here are the key points. ・🎯 Choices, numeric values, and point coordinates are all unified as classification over a finite candidate set ・🧠 No new prediction head is added - it reuses the model's existing query-key attention projections for scoring ・📐 Point prediction uses a 3x3 grid refined through 3 rounds of re-cropping and re-encoding ・🚀 160.6ms average latency on a B200 GPU, 2.5x faster than the best autoregressive baseline and 5.6x faster than the base model ・📊 65.64% average across 12 benchmarks, beating Qwen3-VL-8B and others, ranking #1# on 7 of them ・🔢 Surprisingly, coarser distance bins actually beat finer ones in accuracy ・🌲 A probability-tree decoding scheme extends the single-point predictor to multi-point prediction with no extra training What's interesting is how this shift away from text generation toward a more "intuitive" classification approach delivers both speed and accuracy at once. #VLM# #SpatialReasoning#

TL;DR Treating spatial reasoning as a classification problem over finite candidates, instead of generating text, speeds up a VLM by up to 5.6x while also improving accuracy. Title: MJev-Spatial: System One Spatial Intelligence via Native Attention Matching URL: Here are the key points. ・🎯 Choices, numeric values, and point coordinates are all unified as classification over a finite candidate set ・🧠 No new prediction head is added - it reuses the model's existing query-key attention projections for scoring ・📐 Point prediction uses a 3x3 grid refined through 3 rounds of re-cropping and re-encoding ・🚀 160.6ms average latency on a B200 GPU, 2.5x faster than the best autoregressive baseline and 5.6x faster than the base model ・📊 65.64% average across 12 benchmarks, beating Qwen3-VL-8B and others, ranking #1# on 7 of them ・🔢 Surprisingly, coarser distance bins actually beat finer ones in accuracy ・🌲 A probability-tree decoding scheme extends the single-point predictor to multi-point prediction with no extra training What's interesting is how this shift away from text generation toward a more "intuitive" classification approach delivers both speed and accuracy at once. #VLM# #SpatialReasoning#

Builder’s Pick this week: God’s Eye View 👀 An open-source spatial intelligence project that turns public data into a live 3D view of the world. Aircraft, ships, satellites, earthquakes, traffic, and public cameras can all appear on the same globe, with voice control on top. What makes it interesting isn’t just the interface. It’s the idea of turning scattered public signals into one explorable environment. A very fun repo to poke around if you’re building around maps, geospatial data, visualization, or multimodal agents.

Announcing our partnership with @ManycoreTech Manycore Tech is a leading global provider of spatial intelligence, focused on accelerating the integration of AI into the physical world. Axis Suite delivered a set of complex, physics-ready simulation assets to support Manycore's 3D generative model. A true 3D world model needs more than pixels: it needs interactive environments grounded in physical reality. With these high-fidelity assets integrated into the model, Axis Suite provides the physical signals required to build the next generation of spatial intelligence. Read more about Manycore Tech:

A journey starts as video. At scale, it becomes spatial intelligence. @PathPulse_AI is building a real-time picture of road conditions, movement and safety from video captured across the physical world. As that footprint grows, the data layer behind it has to keep up. As the PathPulse team puts it: “To map the world in real time, you need a powerful foundation. Shelby has been an amazing partner for giving us the reliable, global network we need to stream live road insights and keep our drivers moving safely.” The infrastructure behind spatial intelligence matters as much as the intelligence itself.

AMD’s US$8.2B World Labs deal could be a bigger bet than it looks. The acquisition gives AMD access to spatial intelligence, a key technology behind the emerging physical AI sector. Unlike generative AI that primarily works with text, images and code, physical AI aims to understand how objects and environments behave in the real world. That could open up opportunities across robotics, autonomous systems and industrial simulation, while giving AMD another avenue to compete in the fast-growing AI market. For investors, the key question is whether physical AI can become a meaningful new growth driver for AMD. Stay ahead of the AI trends shaping markets with LongbridgeAI. #LongbridgeSG# #LongbridgeAI# #AMD# #WorldLabs# #FeiFeiLi# #PhysicalAI# #AIInvesting# #ArtificialIntelligence# #Robotics# #TechInvesting# ---- The information provided herein is intended for general informational purposes only and does not constitute investment advice, recommendation, or an offer to buy or sell any investment product or service. All investments carry risks and may not be suitable to everyone. You may lose all or part of your investment principal. Please seek independent financial advice if you are unsure about any investment decisions. This advertisement has not been reviewed by the Monetary Authority of Singapore.

$AMD enters the AI simulation market, challenging $NVDA dominance Atlas is World Labs’ foundation model for spatial intelligence It can take just a few images, reconstruct a persistent 3D world, infer unseen areas, generate new camera viewpoints, and output depth maps, point clouds and Gaussian splats NuRec is NVIDIA’s closest equivalent, used to reconstruct real environments into 3D scenes. Cosmos handles world generation, physical reasoning and robot-action prediction, while Isaac and Omniverse provide the simulation layer World Labs opens AMD to all these new markets, where it was previously essentially a non-player $8.2B may seem expensive, but for a market that could become worth hundreds of billions of dollars within just a few years, the potential ROIC is enormous