Nvidia agreed to pay $12.9 billion for Hugging Face this week, The Information reported. The purchase feels similar to Microsoft's acquisition of GitHub. Nvidia agreed to pay $12.9 billion for Hugging Face this week, The Information reported. The purchase feels similar to Microsoft's acquisition of GitHub.
Nvidia followed developers. It bought the place they already go for AI models. Meanwhile, Anthropic and OpenAI are selling a premium into a market that’s increasingly treating free as the default.
It’s getting easier to run those open models, too. Earlier this week, Ollama shipped a release that lets Claude Desktop run Qwen, DeepSeek, and Kimi. You open Anthropic's app, click the model picker, and pick Kimi K3 instead of Opus 5.
Stop choosing between alert coverage and your budget
Most observability platforms force a tradeoff: alert on everything and pay exponentially, or scale back and accept the blind spots. Join us live on September 10 as we cover two OpenSearch capabilities, PPL and the Unified Alert Manager: no feature tiers, no licensing negotiation, no ingestion ceiling. Here’s what you’ll learn:
How PPL simplifies multi-step, cross-signal alert conditions
How the Unified Alert Manager replaces per-tool notification sprawl
Shopify CEO Tobi Lütke says Claude Code's refusal to read AGENTS.md creates a "complexity tax" for teams managing AI coding agents across large codebases.
Enterprise AI agents are hitting a latency wall as network hops, CPU work and centralized infrastructure push production response times beyond 500ms today.
“The harness is where the hard work is”: Harness bets on agents that enterprises can trust in production
Harness CEO Jyoti Bansal thinks the hard part of AI agents was never building them, it's trusting them in production. His company just launched Autonomous Worker Agents, letting enterprises swap fixed pipeline scripts for agents that deploy, test, and scan under existing governance controls. In this episode, Bansal talks with The New Stack's Frederic Lardinois about why running agents in production is a different beast entirely, and what it takes to make that trust possible.
Alert volumes are climbing, and AI is helping attackers move faster than most SOCs can respond. On September 15, a small group of CISOs and SOC leaders will meet behind closed doors to work out what security operations looks like when AI becomes core to its architecture. Apply now — this room is capped at 25 and built for people who contribute, not an audience that listens.
Now in its 9th year, the AI Infra Summit is the premier event for engineers, architects, and AI/ML practitioners focused on full-stack AI infrastructure. Join 8,000+ attendees to explore cutting-edge hardware, systems, and AI data center innovations. Want to get in on the action at no cost?
More caching and a bigger vector database won't fix what breaks when agent workloads, not human ones, hit your retrieval layer. See why bolted-together pipelines crack under concurrent load — and where this wall shows up in production before you find it yourself.
AI agents write code faster than any team can review it. Pull requests are up nearly 2x, bugs up 54%, and incidents per PR have nearly tripled. Explore where human review still earns its keep — and where a verified pipeline needs to take over.
Cautious Optimism is a newsletter that covers AI models, startups, and the biggest software companies in the world, and then explains what the numbers mean. Here's what shows up in your inbox every morning:
Earnings, interest rates, exits
Where tech meets markets, politics, and the public
"The intelligence of cheaper, smaller models is now much closer to their larger frontier counterparts, providing us a lot of room for cost optimizations."
— Michele Catasta, president and head of AI at Replit
.png?height=72&upscale=true&name=The%20New%20Stack%20logo%20-%20Reversed%20(2).png){kind=small}
