Samwise Tech/AI/Robotics Newsletter
Monday, August 17, 2026
Nvidia Backs $500 Billion AI Data Center Push with GPU Collateral Guarantee
Six of the world's largest investment firms — Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR — have agreed to commit up to $500 billion to build AI data centers, with Nvidia at the center of the deal. The most consequential element: Nvidia has agreed to guarantee with its own funds that its chips used as collateral in these financing arrangements will retain their value. The plan is designed partly to create a secondary market for aging Nvidia GPUs. Bond markets reacted with concern, prompting CEO Jensen Huang to appear on business television and on X to clarify the limits of Nvidia's financial exposure.
Sources: TechCrunch
Anthropic CEO: AI Backlash Is ‘Fundamentally a Crisis of Trust’
Anthropic CEO Dario Amodei pushed back against investor Gavin Baker's claim that his warnings about AI dangers have fueled a backlash against data centers and AI development in the United States. Amodei characterized the phenomenon as “fundamentally a crisis of trust” that has been decades in the making. He disputed Baker's framing, insisting his writing has been “about equally balanced between risks and benefits.” Amodei acknowledged that “by far the most accurate criticism of AI companies including Anthropic is that we haven't yet delivered on our big promises to benefit the world,” and argued that actually curing cancer — not improving messaging — would change public opinion of AI.
Sources: TechCrunch
LG and Nvidia Partner to Build Bipedal Humanoid Robot, Debut Set for Early 2027
South Korean electronics giant LG is developing a next-generation bipedal humanoid robot in partnership with Nvidia, with a public debut planned for the first quarter of 2027. LG Corp chairman and CEO Kwang Mo Koo and Nvidia founder and CEO Jensen Huang signed a memorandum of understanding on August 13, 2026, at Nvidia's headquarters in Santa Clara, California. The robot is powered by Nvidia Jetson Thor for onboard compute and is being built using Nvidia Isaac GR00T and Nvidia Halos for Robotics. The MOU formalizes LG's entry into the humanoid robot market alongside one of the AI industry's most prominent compute providers.
Sources: Robotics and Automation News
DeepSeek Launches Open-Source Agent Harness and Updated V4-Pro Model
Chinese AI lab DeepSeek launched DeepSeek Harness, an open-source agent framework entering developer preview under the MIT license, alongside a general-availability release of its flagship V4-Pro model. Built on a framework called Cordis, Harness uses composable plugins for models, tools, sandboxes, filesystems, orchestration, and user interfaces. V4-Pro runs a 1.6-trillion-parameter architecture with 49 billion parameters active per token and supports context windows up to one million tokens. The smaller V4-Flash uses 284 billion parameters with 13 billion active per token. Beginning August 16, DeepSeek introduced separate peak and off-peak API pricing.
Sources: VentureBeat
North American Robot Orders Reach $622 Million in Q2, Revenue Up 21% Year Over Year
North American robot orders totaled 8,940 units worth $622 million in the second quarter of 2026, up 4.3 percent in unit volume and 21.3 percent in revenue year over year, according to Robotics and Automation News. The figures show automation demand broadening across multiple industries during the April through June period. Revenue grew roughly five times faster than unit volume, pointing to a market increasingly oriented toward higher-cost robotic systems. The Q2 data indicates North American companies across manufacturing, logistics, and other sectors continued to expand their automation deployments through the first half of 2026.
Sources: Robotics and Automation News
Study Finds AI Cannot Yet Fully Automate Open-Ended Scientific Research
A study posted to preprint server arXiv in late July 2026 found that AI systems cannot yet fully automate open-ended scientific research. Two papers submitted to the Neural Information Processing Systems conference were chosen as test cases; an AI tool was asked to conduct research and write papers based on each paper's research question. The original authors reviewed the AI output and were not impressed. “I don't think full automation of open-ended research is on the horizon right now,” said Sayash Kapoor, a Princeton University computer scientist and co-author of the preprint. The line of work was pioneered by Sakana AI, whose AI Scientist system first appeared in 2024.
AI and Self-Driving Labs Could Unlock the Next Semiconductor Materials Breakthrough
The next major semiconductor breakthrough will come from new materials, as conventional options hit quantum tunneling limits, atomic-scale interface effects, and thermal budgets that classical semiconductor materials cannot handle as chip linewidths shrink below two nanometers. An analysis from Robotics and Automation News identifies AI and self-driving labs as the most promising tools for accelerating the discovery of replacement materials. The piece argues that materials science has become the critical bottleneck at the frontier of chip manufacturing, where the physics of established materials can no longer keep pace with industry needs.
Sources: Robotics and Automation News
Tech Pulse
Top Frontier Models: GPT-5.5 (70%) | Claude Opus 4.8 (69%) | Gemini 3.1 Pro (est. top-3)
Top Open Source Models: DeepSeek V4-Pro (leader) | Qwen3-A22B (235B) | Mistral Large 3
Top Small Models (15–50B): Meta Muse Glimmer (30B) | Sarvam-30B | Mistral 3 Small
Top Edge Models (0–15B): Qwen3.5-9B (beats gpt-oss-120B) | Gemini 3.7 Flash | Mistral 3 Mini
AI Leaders: NVIDIA ~$4.6T | Microsoft ~$4T | Alphabet ~$3.25T
Robotics Leaders: Intuitive Surgical (~$165B) | Fanuc (~$35B) | ABB Robotics (spin-off est. ~$20B)
Every newsletter preserved and searchable
Curated by JD · samwise.agency
Leave a Reply
You must be logged in to post a comment.