Semiconductor and AI infrastructure company reorganizing beyond GPU sales into broader platform businesses.
- Private markets vs. the S&P 500
- Causal Analysis of Robot Foundation Models Enables Efficiency Gains
- A Phenomenology of Neural Geometry in Vision-Language-Action Models
- 🤝 An AI truce
- Investors Turn on Data Center Builders as Rising Interest Rates Dampen the AI Trade
- Higgsfield Hit $1B in Annualized Revenue in 18 Months. Cursor Needed 24. Here Are Alex Mashrabov’s 10 Calls
- Whole-Body Dexterity
- The tech vertical taking off in Europe
- 💰 Anthropic's money bags
- The company building underwater drones for $50,000
- Thrive Capital's Bet on Neuralink, OpenAI & the Future of Medicine
- Modeling and Understanding Neural Weights of Humanoid Policies with Diffusion
“Pigment... $250M Series E at $2.3B, Nvidia also plans to invest.”
“Inference chips are different from GPUs in that they're specifically built to answer questions asked of an AI model. They are designed to be faster and cheaper than GPUs.”
“Scaleup Europe Fund led, joined by World Innovation Labs, Greenoaks, Iconiq, IVP, Nvidia, and Meritech.”
“DoubleAI, a three-year-old Jerusalem startup that develops AI models that reduce computing costs by optimizing how software runs on CPUs and GPUs, raised a $220 million Series A round at a $1+ billion post-money valuation. The deal was co-led by Lightspeed Venture Partners and Bessemer Venture Partners, with NVIDIA, Dell Technologies Capital...”
“Microsoft pulled out of a deal with Nvidia-backed Nscale for a UK-based data center due to delays with the electricity grid supplying power to the site.”
Source→“latency dropped from 311 ms to 209 ms, and throughput increased by 44% on an H100”
Source→“DoubleAI, an Israeli startup focused on AI cost containment, raised $220m from Lightspeed, Bessemer Venture Partners, Nvidia, and Dell.”
“Microsoft pulled out of a deal with Nvidia-backed Nscale for a UK-based data center due to delays with the electricity grid supplying power to the site.”
Source→“Reflex Aerospace, a German maker of satellite platforms, raised funding from Nvidia.”
“Cosmos3 'treats action as a first-class modality with per-domain action encoders'”
Source→“the program's $10M expansion with NVIDIA and Accel”
“Nvidia-backed AI cloud provider Firmus shelved an IPO that sought a valuation north of $30 billion.”
Source→“Google DeepMind and NVIDIA, Co-developers of MuJoCo Warp (GPU-accelerated MuJoCo), Referenced as the GPU simulation backend used for benchmarking collision methods”
Source→“That could give companies more alternatives to Nvidia GPUs and proprietary models from leading AI companies, potentially shifting some of the economics of AI.”
Source→“Nemotron 3.5 Super VL — NVIDIA’s open multimodal model for specialized agents (77 votes)”
Source→“NVIDIA, GPU hardware, RTX 5090 used for GPU simulation benchmarks (Appendix B, Table 4)”
Source→“turning automotive 'into a computing industry, a technology industry'”
Source→“kind of like AWS for today, and so it's just not AWS for the future”
Source→“runs at 50Hz on an NVIDIA RTX 5090 on the operator workstation, with 11.8ms median inference time (17.3ms 95th percentile) and 50.1ms median state-to-command delay”
Source→“Meta Muse is doing fantastic... it's vastly NVIDIA. And they were zero last year”
Source→“He cited Nvidia's OpenShell, with 'over 100 companies now joining us'”
Source→“Lilly winning, Walmart winning, FedEx winning... Boeing winning, the US military winning”
Source→“we generate references from fine-tuned GR00T N1.7-3B VLAs [17]”
Source→“anything you can spend to deflect any anti-Americanism is probably worth the money.”
Source→“We built a fully interactive one-bedroom apartment in NVIDIA Isaac Sim [114] through a reproducible pipeline”
Source→“Nvidia CEO Jensen Huang said he'd cut off access to chip supplies for anyone developing unsafe AI.”
Source→“a great enterprise businesses through Dell and HP and Lenovo and others”
Source→“SPICE is still going to be useful, and Verilog will be useful, and simulators will be useful... Photoshop, and AutoCAD, and Dassault, CATIA”
Source→“You can't separate capability, functionality from safety and security. They go as one.”
Source→“Salesforce announced they've got their own models that are built on Nemotron and built with NVIDIA.”
Source→“Contrary to prior belief, discrete action parametrizations, even discrete regression (DR) with the simplest bin discretization, can exhibit greater control precision than their continuous counterparts, despite the possibly lossy quantization of actions!”
Source→“NVIDIA et al., 2025 cited among VLAs that predict actions using continuous-space generative models”
Source→“discrete flow is competitive with, and often slightly surpasses, continuous flow. More surprisingly, discrete regression consistently outperforms continuous regression.”
Source→“The simulation framework mjlab is MuJoCo-based but GPU-accelerated, and training uses 4,096 parallel environments”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.