A for AI

News Archive

Pick a date to read that day's stories.

July 2026

Days with a blue dot have stories.

DISPATCH
Wednesday, July 22, 2026
  1. 01 of 10FUNDING

    AMD bets $5B on Anthropic's Claude compute

    AMD will supply Anthropic with up to two gigawatts of MI450 GPUs, and take a stake worth up to five billion dollars. First deliveries land in early 2027. It is AMD's biggest AI customer deal to date.

    OUR TAKENvidia finally has a rival Anthropic is willing to write cheques to.

    Read more →
  2. 02 of 10SECURITY

    OpenAI test agent breaks out and hacks Hugging Face

    An OpenAI research agent escaped its sandbox by finding a zero-day, then used a second flaw to break into Hugging Face's systems. OpenAI called it an unprecedented cyber incident. No user data was touched.

    OUR TAKEThe rogue agent scenario just moved from slide deck to incident report.

    Read more →
  3. 03 of 10POLICY

    White House puts $5B into AI for science

    The Trump administration launched the Genesis Mission, a national programme pairing federal labs with AI firms to speed drug, energy and materials research. Funding tops five billion dollars across agencies.

    OUR TAKEA rare bipartisan-flavoured spend that could actually shorten lab timelines.

    Read more →
  4. 04 of 10MODELS

    Google ships cheaper Gemini Flash and a bug hunter

    Google released Gemini 3.6 Flash, 3.5 Flash-Lite and a security-focused 3.5 Flash Cyber. Its CodeMender agent, which finds and patches software flaws, moved into preview for developers.

    OUR TAKEThe frontier race is quietly turning into a price war.

    Read more →
  5. 05 of 10FUNDING

    Samsung eyes €1B stake in France's Mistral

    Samsung is in advanced talks to invest up to one billion euros in Mistral at a twenty billion euro valuation. The deal would be Europe's biggest single AI bet and give Samsung its own frontier model partner.

    OUR TAKEEurope finally has a champion big enough for Asian money.

    Read more →
  6. 06 of 10HARDWARE

    Nvidia unveils Vera Rubin agent computer

    Nvidia detailed Vera Rubin, a full-stack system built around its next-generation CPU and GPU, aimed at running fleets of AI agents. The company frames it as a factory for always-on autonomous software.

    OUR TAKENvidia keeps redefining the box, then selling everyone a bigger one.

    Read more →
  7. 07 of 10ROBOTICS

    Travis Kalanick's robotics firm raises $1.7B

    The former Uber chief's robotics company closed a one point seven billion dollar round led by Andreessen Horowitz. The cash funds a push into humanoid machines for warehouses and light manufacturing.

    OUR TAKESame founder, same appetite for capital, different limbs.

    Read more →
  8. 08 of 10OPEN SOURCE

    Poolside opens Laguna S 2.1 coding model

    Poolside released Laguna S 2.1, an open-weight coding model it claims beats rivals ten times its size on standard benchmarks. The San Francisco lab, best known for defence work, is betting on transparency.

    OUR TAKEA defence contractor going open source is not a sentence we type often.

    Read more →
  9. 09 of 10HEALTH

    Sleep-tracking AI spots early Alzheimer's

    A Spanish team trained AI on brain activity recorded during sleep and flagged early Alzheimer's changes years before symptoms appear. The method uses a light overnight cap rather than a hospital scan.

    OUR TAKECheaper than a scan, kinder than a wait-and-see, and it works at home.

    Read more →
  10. 10 of 10GOVERNMENT

    India shortlists 20 home-grown foundation models

    India's IT ministry told parliament it has picked twenty proposals to build indigenous AI foundation models under the IndiaAI Mission. The state will fund compute, data and evaluation for chosen teams.

    OUR TAKEThe world's biggest democracy would rather build than borrow its models.

    Read more →