Why Every AI Lab Is Now Racing to Build Agents

In Brief

One product, Claude Code, triggered a strategic chain reaction across every major AI lab. The race has shifted from building the smartest model to building the most useful autonomous agent, and enterprise customers are the prize.

What Happened

Every major AI lab just tipped its hand. In a single week, OpenAI, Anthropic, Google, Meta, and xAI all revealed how dramatically their strategies have shifted toward autonomous AI agents built for business.

Reuters reported that OpenAI is pursuing partnerships with private equity firms including TPG, Bain Capital, and Brookfield in deals potentially worth $10 billion. The logic: PE firms control massive portfolios of companies and influence their tech spending. OpenAI also announced it is folding ChatGPT, its Codex coding tool, and its Atlas browser into a single desktop "superapp." And they are throwing everything into building a fully automated AI researcher, with plans for an autonomous "AI research intern" by September.

The competitive pressure is real. According to enterprise software vendor Ramp, business customers buying AI for the first time are choosing Anthropic at three times the rate of OpenAI. That is a reversal from a year ago. Anthropic is also reportedly courting its own PE partnerships with Blackstone and others.

SmarterX founder and CEO Paul Roetzer traced the entire shift back to a single inflection point on Episode 205 of The Artificial Intelligence Show.

The Key Numbers

• 3-to-1 rate at which new enterprise customers are choosing Anthropic over OpenAI

• $10 billion in potential PE-backed enterprise deals OpenAI is pursuing

• 700 experiments run autonomously by Andrej Karpathy's AI research agent in two days

Why Claude Code Changed Everything

The shift started on January 6. That is when Roetzer first flagged what was happening. "This goes back to Episode 189 of the podcast on January 6th," he says. "That was when Claude Code sort of blew up, and it became very hot over those last two weeks of 2025. Something had definitely changed."

Claude Code is an AI tool that writes and edits software autonomously. Other labs had similar tools, but this one worked at a level that got the entire industry's attention. "They did something different with the harness, like how they enabled it to do what it does," says Roetzer. "All these labs see not the finish line, but like the next mile marker of AGI agentic capability."

Google's own engineers said it out loud. Roetzer pointed to a tweet from Jaana Dogan, a principal engineer at Google, who wrote on January 2nd: "I gave Claude Code a description of the problem, it generated what we built last year in an hour." Google DeepMind's Logan Kilpatrick posted that "all the industries you thought weren't going to be disrupted by AI are about to be disrupted." The tweet was quickly deleted.

The autonomous race is accelerating. Former OpenAI founding member Andrej Karpathy demonstrated what analysts are calling the "Karpathy Loop," an AI agent that ran continuously for two days, executed 700 experiments, and improved language model training by 11%. Shopify CEO Tobias Lutke replicated the approach overnight and achieved a 19% performance gain. Karpathy called autonomous research "the final boss battle" for AI labs.

"It is an all out race for agents and they're seeing a pot of gold with enterprise adoption, which is why Anthropic and OpenAI are doing deals with PE firms. It's why they're doing alliances with major consulting firms. They're trying to get in and get where this is going to be because the labor replacement value of being the model they go to when they reduce workforces and put it all into AI models to token max to get work done, they see that future coming very fast."

Paul Roetzer, founder and CEO of SmarterX, Episode 205 of The Artificial Intelligence Show

The fallout is everywhere. Meta delayed its latest AI model after it fell short of rivals. Elon Musk admitted xAI "was not built right first time around." Microsoft restructured its AI leadership after Copilot drew just 6 million daily users compared to ChatGPT's 440 million.

"It's very, very important that you understand what we just covered," Roetzer says. "That's what these labs are doing and it's going to become very apparent I think in the next three to six months that this is full go where they're headed."

SmarterX Take

The AI labs have collectively decided that autonomous agents deployed inside businesses represent the largest near-term opportunity in AI. That is why they are simultaneously courting private equity, restructuring products, and racing to automate their own research.

For business leaders, the competitive dynamics are actually favorable right now. Both OpenAI and Anthropic are aggressively pursuing partnerships, offering credits and forward-deployed engineers. Companies that engage now have leverage they will not have in 12 months when pricing power shifts back to the labs.

What to Watch

OpenAI's September deadline for its "AI research intern" is the first concrete timeline. If they deliver an autonomous system that solves specific research problems independently, it validates the broader thesis that these agents are months from transforming business workflows.

Anthropic's enterprise lead may be the most important number in AI right now. Winning 70% of head-to-head enterprise comparisons is commanding, but OpenAI has distribution advantages through Microsoft. How those two split the enterprise market over the next six months will shape the industry.

Further Reading

OpenAI Courts Private Equity to Join Enterprise AI Venture → reuters.com

OpenAI Plans Launch of Desktop Superapp → wsj.com

OpenAI Is Throwing Everything into Building a Fully Automated Researcher → technologyreview.com

Anthropic Turns the Tables on OpenAI in Critical Revenue Category → axios.com

Why Everyone Is Talking About Andrej Karpathy's Autonomous AI Research Agent → fortune.com

Heard on The Artificial Intelligence Show, Episode 205
Paul Roetzer and Mike Kaput break down the strategic shift happening across every major AI lab and what the race for autonomous enterprise agents means for business leaders. Listen →