Industry·1 min read·NVIDIA

NVIDIA Reveals H200 Chip: Double the AI Power

NVIDIA unveiled the H200, a new AI chip that doubles performance for training and inference of large models.

NVIDIA Reveals H200 Chip: Double the AI Power
Share:

H200: The Next Generation of AI Chips

NVIDIA revealed the H200 chip, offering double the performance of the H100 for training large language models. The new chip includes 141GB of HBM3e memory and 4.8TB/s of bandwidth.

The H200 is expected to be available in Q1 2025, with massive demand from leading AI companies.

Want AI news before everyone else?

The morning's most important AI stories, straight to your inbox. No fluff.

Related Articles

RUST-LANG/RUST · LLM POLICY Review it. Don’t write it. BANNED LLM-authored code Unmarked LLM comments Pasted review replies Soundness-critical work WITH DISCLOSURE Machine translation Trivial code or prose LLM-assisted bug finds Pre-arranged patches NO DISCLOSURE Asking questions Analysing and refining Reviewing other PRs Output only you see Five teams · the rust-lang/rust monorepo · adopted 5 August 2026 BITSMINDS.COM
Industry

Rust Bans LLM-Written Code in Its Compiler Monorepo

NINTH CIRCUIT · AUGUST 4, 2026 “Perplexity does not access. Users do.” USER types the request COMET AGENT clicks, scrolls, buys AMAZON.COM serves the page INJUNCTION VACATED access prong only · remanded to San Francisco BITSMINDS.COM
Industry

Perplexity Wins: Its Agent Doesn't Access Amazon, You Do

WHITE HOUSE — VOLUNTARY CYBER EVALUATION OpenAI agent breached Hugging Face Anthropic models breached 3 companies Google attending Meta attending Submit up to 30 days before public release TESTS FINALIZED Announced Monday, Aug 3 METRICS USED HOW RESULTS REPORTED WHETHER ANY IS PUBLIC Details withheld BITSMINDS.COM
Industry

White House Finalizes AI Cyber Tests After Agent Breaches