Independent reporting on artificial intelligence
Edition No. 16
AI, closely read
THE CONTEXTEER
Wednesday
October 28, 2026
The daily editionWednesday, October 28, 2026
No. 16 · 13 stories · Archive
Compute & Infrastructure

Google gains first outside hyperscaler for TPUs as Oracle plans offering

Oracle will also add wafer-scale accelerators from Cerebras Systems and OpenAI's Jalapeño inference processor, and plans to spend $70B on capital projects in the year to May 2027.

Oracle plans to offer Google's TPU artificial intelligence accelerators through its public cloud starting next year, the first time Google has supplied the chips to another hyperscaler, SiliconANGLE reported on Oct. 27. Oracle will also add wafer-scale accelerators from Cerebras Systems and OpenAI's Jalapeño inference processor, CNBC reported, according to SiliconANGLE.

Read the full story
Mosaic of hand-cut ceramic tiles showing a bustling covered market where shoppers buy small chips while one whole round wafer sits on a pedestal, in fuchsia, emerald and saffron on navy.
Image courtesy of Fay Tograph · AI photographer, made with Grok image generation
  1. 01

    DeepMind starts rolling out Genie 4 world model to trusted testers

    Genie 4 keeps a scene consistent for up to two minutes, double the limit of Genie 3, though details can still drift the longer it runs.

  2. 02

    Perturbation-based LLM ambiguity detection beats baselines, authors report

    On four datasets, the authors report, the methods beat baselines that measure uncertainty in the model's answers or rewrite prompts with model paraphrases.

  3. 03

    NVIDIA Vera Rubin NVL72 systems now generally available on Oracle Cloud

    Each rack connects 72 GPUs, while Vera CPUs handle the work agents do between model calls, such as executing tools, testing code and retrieving data.

  4. 04

    Amazon calls AI tool adoption metrics for engineers deeply flawed

    In a memo urging engineers to learn and apply AI, Amazon says it is best to be measured by long-term outcomes, while acknowledging they are difficult to measure.

  5. 05

    Faster-AI groups spend roughly $8 million, safeguard backers $6.9 million

    Leading the Future, a network backed by Andreessen Horowitz and OpenAI President Greg Brockman, launched in August 2025 with more than $100 million in commitments.

  6. 06

    Bank of America warns customers not to give banking log-ins to AI agents

    A Bank of America spokesperson confirmed the email was authentic; it said customers are responsible for activity resulting from such access.

  7. 07

    Anthropic agreement gives government buyers 40% off Claude

    The discount covers direct purchases from Oct. 26 and cloud-provider purchases in the coming months.

Business · Update

Nvidia and AMD to invest $4 billion in Meta

Meta's $95 billion raise is structured as an initial $35 billion in underwritten public offerings and an additional $60 billion through an at-the-market program.

Today's edition has 5 stories, 7 briefs and 1 update. 275 claims were checked against the cited sources; 2 stories were cut because they did not pass.