Samwise Tech/AI/Robotics Newsletter
Sunday, June 7, 2026
Anthropic Files Confidential S-1, Targets $965 Billion IPO
Anthropic filed a confidential draft S-1 registration statement with the U.S. Securities and Exchange Commission on June 1, 2026, beginning the formal process toward an initial public offering. No share count or price range was set at filing. The move followed a $65 billion Series H round completed days earlier, pushing Anthropic’s post-money valuation to $965 billion — the highest for any private AI company. Revenue run-rate reached approximately $47 billion in May 2026, up from roughly $10 billion the prior year. Anthropic has committed to paying SpaceX $1.25 billion per month through May 2029 for computing infrastructure.
Sources: TechCrunch
Stanford’s 2026 AI Index: Record Investment, Accelerating Models, Falling Regulatory Trust
The 2026 Stanford AI Index from Stanford University’s Human-Centered Artificial Intelligence center documents three simultaneous trends: accelerating capabilities, record investment, and declining regulatory trust. U.S. organizations released 50 notable AI models in 2025, with China closing the gap. Global AI compute has grown roughly three-fold annually since 2022, reaching 30 times 2021 levels. Nvidia accounts for over 60 percent of world AI compute. AI investment hit a record $581 billion in 2025, more than double 2024. Just 31 percent of U.S. survey respondents said they trusted the government to regulate AI — the lowest confidence of any country measured.
Sources: IEEE Spectrum
Nvidia RTX Spark Superchip Targets $200 Billion PC Market with On-Device AI Agents
Nvidia announced RTX Spark at Computex 2026 in Taipei on June 1, an Arm-based superchip the company calls a bid to reinvent the Windows PC for the age of AI agents. The chip integrates a Blackwell GPU with 6,144 CUDA cores, up to 20 Arm CPU cores, and 128GB of LPDDR5X unified memory, delivering one petaflop of AI compute. RTX Spark can run 120-billion-parameter language models locally with one-million-token context windows. Consumer PCs using RTX Spark arrive this fall from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI. Adobe is rebuilding Photoshop and Premiere Pro around the RTX Spark architecture.
Sources: TechCrunch
Hello Robot Ships Stretch 4 Home Robot at $29,950 — First Run Sold Out
Hello Robot released Stretch 4, the fourth generation of its home-assistance robot, at $29,950 and available now. The flagship upgrade is an omnidirectional wheeled base allowing movement in any direction without rotating first — a design the company calls notoriously difficult to engineer. Founded in 2017 by former Google robotics director Aaron Edsinger and Georgia Tech professor Charlie Kemp, Hello Robot designs Stretch as an open-source platform for Physical AI research and general-purpose robotics. The robot ships in a standard cardboard box via UPS or DHL. The initial production run of 200 to 300 units, made at Hello Robot’s Martinez, California headquarters, sold out immediately.
Sources: TechCrunch
MiniMax M3 Launches as Open-Weight Frontier Model at 5–10% of Proprietary Cost
MiniMax released its M3 large language model on June 1, 2026, combining frontier-tier coding and agentic performance with a one-million-token context window and native multimodality at a fraction of leading proprietary model costs. On SWE-Bench Pro, M3 scores 59.0 percent, outperforming GPT-5.5 and Gemini 3.1 Pro. At maximum context, per-token compute demand falls to one-twentieth of the prior generation, with nine-fold prefill and fifteen-fold decoding speed gains. Introductory API pricing starts at $0.30 per million input tokens. MiniMax announced plans to release open weights within ten days, making M3 available for enterprise download and customization.
Sources: VentureBeat
U.K. Regulators Force Google to Let Publishers Opt Out of AI Search — A World First
Google announced on June 3, 2026, compliance with requirements from the U.K.’s Competition and Markets Authority, which the CMA called a world first: publishers may now use a toggle in Google’s Search Console to opt their sites out of AI Overviews, AI Mode, and AI Overviews in Discover. The CMA designated Google as having strategic market status last October and directed in January that publishers gain control over their content’s use in AI features. The opt-out will be tested with U.K. publishers before a global rollout. Google confirmed that an opt-out decision will not affect a site’s traditional search rankings.
Sources: TechCrunch
Perplexity Unveils Hybrid Local-Cloud AI Inference at Computex, Launching in July
Perplexity AI unveiled what it calls the first hybrid local-server inference orchestrator at Computex 2026 in Taipei, with CEO Aravind Srinivas demonstrating the system alongside Intel CEO Lip-Bu Tan during Intel’s keynote on June 2, 2026. The software decides in real time which parts of an AI query are handled on the user’s local device and which are routed to cloud-based frontier models. The demonstration ran on Intel Core Ultra Series 3 processors using Perplexity’s Personal Computer agent. The hybrid inference feature will launch for Perplexity Computer users on Windows in July. Perplexity has raised over $200 million at a $20 billion valuation.
Sources: VentureBeat
Tech Pulse
Top Frontier Models (SWE-Bench): Claude Mythos Preview (93.9%) | GPT-5.5 (82.6%) | Gemini 3.1 Pro (78.8%)
Top Open Source Models (SWE-Bench Pro): MiniMax M3 (59.0%) | Kimi K2.6 (58.6%) | DeepSeek V4-Pro (80.6% Verified)
Top Small Models (15–50B): MAI-Thinking-1 (35B) | Qwen3-32B | Mistral Medium 3 (22B)
Top Edge Models (0–15B): Phi-4 (14B) | Gemma 3 9B | Qwen3-8B
AI Leaders (mkt cap): Nvidia $4.97T | Alphabet $4.63T | Microsoft $3.11T
Robotics Leaders (mkt cap): Intuitive Surgical $175B | ABB $165B | Figure AI $2.6B (private est.)
Curated by JD · samwise.agency
Leave a Reply
You must be logged in to post a comment.