Anthropic and Elon Musk cornered Sam Altman this week

Weekly Update | Issue 519

Together with

How Anthropic and Elon Musk cornered Sam Altman this week

Welcome, TNS Daily readers! We have added this Saturday newsletter, which focuses on the breakneck world of AI and software development and is written by Matt Burns, Chief Content Officer at Insight Media Group. We hope you enjoy this weekend's long read. Scroll to the end of this email for our quote of the week.

The AI race used to be about who had the best model. This week, it looked more like a fight over power plants, data centers, and who gets access to the GPUs first. Anthropic’s new SpaceX deal doubled Claude Code's power-user limits, but that’s the small story. The bigger story is that Anthropic just rented the full computing power of Elon Musk’s Colossus 1 facility in Memphis. Meanwhile, Musk and Sam Altman were in federal court this week, where Musk is trying to unwind OpenAI's transformation into a for-profit company.

Let’s remember the larger picture: Elon Musk co-founded OpenAI with Sam Altman, and Anthropic's CEO, Dario Amodei, was a very early employee at OpenAI as its VP of Research. Both Musk and Amodei split from OpenAI, built rival labs, and traded public criticism themselves. Now they’re partnered on a common cause plaguing every frontier AI company, including OpenAI: compute.

As Alex Wilhelm writes in Cautious Optimism this week, I thought Elon hated Anthropic. Maybe he does, but it’s also possible Musk hates Altman even more because on Wednesday, he rented Anthropic all the compute capacity at the facility he built for xAI to train Grok.

Musk and Altman’s trial is the show. The SpaceX deal is the move.

Read the full story here →

— Matt Burns, Chief Content Officer

Together with FinOut

Your cloud bill shouldn't need a translator

Cloud, Kubernetes, and AI spend is variable, shared, and impossible to explain with legacy tools.
Finout unifies every cost source, from AWS to Anthropic, with Virtual Tags that deliver 100% allocation and zero manual tagging.
Finance and engineering, finally aligned.

See how it works

How Microsoft is governing thousands of Kubernetes clusters without manual intervention

In this edition of The New Stack Makers, Stephane Erbrech, principal software engineer at Microsoft, tells The New Stack how Microsoft Azure Kubernetes Fleet Manager and Cilium Cluster Mesh tackle the complexity of managing thousands of Kubernetes clusters across distributed environments at scale.

Catch the episode

Essential reads

The introverts’ edge: How AI is leveling the developer floor
AI coding assistants aren't just autocomplete anymore — they're quietly reshaping how junior developers learn, survive, and advance in an industry that was already failing them.

I tested the new OpenAI Codex features on a real Python codebase, and it’s the strongest Claude Code rival yet
OpenAI has updated Codex with computer use, an in-app browser, and PR reviews. Testing on HTTPie revealed how these features handle real-world Python bug fixes.

The context window has been shattered: Subquadratic debuts a 12-million-token window
Subquadratic has launched a new AI architecture featuring a 12-million-token context window that outperforms GPT-5.5 on retrieval benchmarks.

What the 2026 State of Open Source report reveals about digital autonomy
The 2026 State of Open Source report highlights the shift toward digital autonomy, balancing vendor freedom with maintenance and security.

How to find and unlock the data hidden within videos
Unlock video data with hybrid search. Learn to use Vespa and CLIP for efficient video preprocessing and multimodal retrieval.

Featured webinars

May 12

Beyond the SRE: Democratizing observability data with GenAI

GenAI is transforming observability from a specialized SRE domain into a democratized asset, allowing anyone to query complex system data using plain language. Join us on May 12 to democratize insights via natural language, bridge data silos with OpenTelemetry, and future-proof your platform for proactive, agentic AI.

Register today →

On demand

Your delivery pipeline’s impact on AI productivity

Companies are pouring money into AI tools, yet software delivery isn’t getting any faster. Why? Because AI-generated code is being throttled by manual pipelines, security hurdles, and outdated deployment processes. Learn why your delivery foundation matters more than your AI model.

Watch on-demand →

On demand

From vectors to tensors: Expanding the possibilities of AI search

Vector embeddings changed search, but they represent only a single point in space. For complex, multimodal, or highly nuanced ranking, you need more dimensions. You need tensors. Catch the replay for a concise, practical primer on why tensors are the next leap for production AI.

Watch on-demand →

On demand

Break the cycle of spending $250bn on manual IT operations

In this deep dive, we took a hard look at why rules-based automation has stalled out, solving only 20% of IT problems while leaving the other 80% to be handled by expensive, manual "stitching."

Watch on-demand →

From our sister site, roadmap.sh

Now live: The LeetCode roadmap

Learning how to solve technical interview problems can feel disorganized, but this LeetCode roadmap helps group things into a logical sequence. It’s a guide that moves away from random "grinding" and instead organizes LeetCode challenges into a logical sequence of data structures and algorithmic patterns.

Explore the roadmap

Quote of the week

"The shift changes workflows from cautious prompt budgeting to deeper reasoning, bigger tasks, and more complete engineering output."

— Elmer Morales, founder of koderAI, on the new Anthropic and SpaceX partnership that increases Anthropic's compute capacity via 220,000 SpaceX-owned Nvidia GPUs. Read more →

Connect with The New Stack