Samwise Tech/AI/Robotics Newsletter
Sunday, May 10, 2026
AI Handles the Work as Cloudflare Cuts 1,100 Jobs Despite Record Revenue
Cloudflare announced it is eliminating 1,100 positions — roughly 20 percent of its global workforce — citing artificial intelligence efficiency gains that rendered the roles obsolete, even as the company posted its strongest quarter on record. CEO Matthew Prince said internal AI usage surged more than 600 percent in just three months, allowing the San Francisco-based security and networking firm to handle greater workloads with fewer employees. Revenue for Q1 2026 reached $639.8 million, a 34 percent year-over-year rise. The layoffs mark the first mass workforce reduction in Cloudflare’s 16-year history and echo similar moves at Meta, Microsoft, and Amazon in recent months.
Sources: TechCrunch
Nvidia Crosses $40 Billion in AI Equity Bets as Chip Strategy Expands
Nvidia has committed more than $40 billion to equity investments in AI companies in the first months of 2026, cementing its strategy to back the broader AI ecosystem beyond selling chips. The company struck agreements this week alone with data center operator IREN — rights to invest up to $2.1 billion — and Corning, with rights to invest up to $3.2 billion. The total eclipses Nvidia’s entire venture activity from prior years combined. The chipmaker reported fiscal Q4 2026 revenue of $39.3 billion, yet analysts say the equity push signals Jensen Huang’s intent to build dependency and loyalty across every layer of AI infrastructure.
Sources: TechCrunch
Anthropic’s Mythos Uncovers 423 Firefox Security Bugs in One Month
Anthropic’s Mythos model helped Mozilla identify and patch 423 Firefox security bugs in a single month — compared to just 31 fixes over the same period a year earlier — according to Mozilla engineers. The model uncovered vulnerabilities dormant in the codebase for over a decade, including several rated high-severity. Anthropic had warned at Mythos’s April launch that its flaw-finding capability was so advanced that thousands of issues needed remediation before public release. Mozilla’s team still relies on human engineers for deployable patches; Mythos-generated code serves as a reference model rather than a direct production fix.
Sources: TechCrunch
OpenAI Releases Three Real-Time Audio Models for Voice Agent Development
OpenAI has expanded its developer API with three new real-time audio models: GPT-Realtime-2 for conversational task execution, GPT-Realtime-Translate for multilingual speech translation across more than 70 languages, and GPT-Realtime-Whisper for live transcription and captioning. The tools enable developers to build voice agents capable of interrupting a speaker mid-sentence, maintaining coherent multi-turn conversations, and switching languages in real time. Pricing starts at $0.40 per 1,000 input audio tokens. The release targets customer service, telehealth, and educational platforms requiring low-latency spoken interaction, and follows OpenAI’s earlier expansion of its image-generation API into enterprise tiers.
Sources: TechCrunch
Apple and Intel Reach Preliminary Deal to Manufacture Future Chips
Apple and Intel have reached a preliminary agreement for Intel to manufacture some future Apple chips, ending the companies’ manufacturing estrangement and giving Intel a high-profile anchor customer for its advanced 18A fabrication process. Intel shares jumped 14 percent on the news. Terms were not disclosed, but analysts believe the contract targets M-class chips for Macs and iPads, potentially starting on Intel’s 18A-P node in 2027. Apple currently relies solely on TSMC for its silicon, and the agreement aims to diversify supply as AI chip demand strains TSMC capacity. The U.S. Commerce Department played a role in facilitating negotiations between the two companies.
Sources: TechCrunch
China’s Moonshot AI Raises $2 Billion at $20 Billion Valuation on Open-Source Demand
China’s Moonshot AI has raised $2 billion at a $20 billion valuation, making it one of the most valuable open-source AI companies globally. The round was led by a consortium of Chinese sovereign-linked funds and international investors. Moonshot’s Kimi K2.5 model currently tops open-source leaderboards, with a 99.0 percent score on HumanEval and 92.0 percent on MMLU. Demand for open-weight models has accelerated as enterprises seek alternatives to subscription-gated frontier systems. The funding underscores a global race in open-source AI, with Chinese labs increasingly competitive against U.S. counterparts on flagship benchmark evaluations.
Sources: TechCrunch
Hackers Deface Canvas School Login Pages, Threaten to Publish Student Data
Hackers defaced several school login pages on Instructure’s Canvas platform, threatening to release stolen student data by May 12 unless the company “negotiates a settlement.” The extortion follows a breach disclosed May 5, in which the same group obtained student names, email addresses, phone numbers, and private teacher-student messages from at least two U.S. institutions. Instructure took Canvas offline after discovering the defacements, calling the action a precaution. The attackers exploited a vulnerability in the Free-For-Teacher account tier. Canvas serves more than 30 million users across thousands of academic institutions. Instructure said it is working with law enforcement and external security researchers.
Sources: TechCrunch
Tech Pulse
Top Frontier Models (SWE-bench): Claude Mythos Preview (93.9%) | Claude Opus 4.7 (87.6%) | GPT-5.5 (~82%)
Top Open Source Models (Arena Elo): GLM-5 / Z.AI (1451) | Kimi K2.5 / Moonshot (1451) | MiniMax M2.5 (1446)
Top Small Models (15–50B): NVIDIA Llama Nemotron 49B | Qwen3-30B-A3B | ServiceNow Apriel-15B
Top Edge Models (0–15B): Gemma 3 12B | Phi-4 14B | Qwen3-8B
AI Leaders: NVIDIA $5.2T | Alphabet $4.2T | Microsoft $3.2T
Robotics Leaders: Intuitive Surgical $175B | ABB Robotics $165B | Boston Dynamics (private est.)
Curated by JD · samwise.agency