AI News for February 22, 2026 — 7min.ai

[NEWS] OpenAI raises cash burn forecast by $111B, won't turn cash-flow positive until 2030: OpenAI now expects cumulative cash burn of $665B through 2030, roughly $111B more than previously forecast.
[NEWS] Altman warns "the world is not prepared" as OpenAI uses AI to accelerate its own research: OpenAI CEO Sam Altman said AGI is "pretty close" and superintelligence "not that far off," revealing that internal models are speeding up development faster than expected.
[TOOLS] Anthropic adds desktop preview, auto-review, and PR merge features to Claude Code: Claude Code can now spin up dev servers and display running web apps directly in the interface, spot errors, and fix them autonomously.
[NEWS] Microsoft discovers "AI Recommendation Poisoning" hidden in Summarize buttons across 31 companies: Microsoft's Defender Security Research Team found that dozens of companies embed hidden prompt injections inside innocent-looking "Summarize with AI" buttons.
[NEWS] Altman calls space data centers "ridiculous," says AI water usage claims are "totally fake": Sam Altman dismissed Elon Musk's orbital data center ambitions at an India event, calling the idea "ridiculous" given current launch costs and the difficulty of repairing chips in space.
[NEWS] China defies global AI scare trade as investors chase winners instead of fleeing disruption: While US markets dump software firms and wealth managers over AI disruption fears, Chinese investors are doing the opposite: chasing perceived AI winners drawn by growth prospects and cost-saving potential.
[OPINION] Google VP warns LLM wrappers and AI aggregators have their "check engine light" on: Google's VP of global startups says two once-hot AI startup models are running out of runway: thin LLM wrappers and model aggregators that route between APIs without proprietary tech.
[NEWS] US farmers and communities reject multimillion-dollar data center bids nationwide: Farmers across Kentucky, Pennsylvania, and Wisconsin are turning down offers worth tens of millions for their land as tech companies scramble to build AI data centers.
[TOOLS] NTransformer runs Llama 70B on a single RTX 3090 via NVMe-to-GPU streaming: A new open-source C++/CUDA inference engine called NTransformer runs Llama 3.1 70B on a single RTX 3090 (24GB VRAM) by streaming model layers through GPU memory via PCIe, with optional NVMe direct I/O that bypasses the CPU entirely.
[OPINION] PromptQL CEO: AI job doomsday predictions are Silicon Valley narcissism: PromptQL's CEO says dire AI job predictions from figures like Mustafa Suleyman and Matt Shumer are a case of Valley self-projection.

About 7min.ai · Subscribe to newsletter