The AI Cold War Is Heating Up

The Cold War in artificial intelligence has finally burst out into the open.

For the past two years, the rivalry between OpenAI, the company behind ChatGPT, and Anthropic, the maker of Claude AI, stayed mostly confined to model releases, benchmark scores and developer tools.

But today the animosity between the two AI giants has become impossible to ignore.

In Washington, OpenAI shot past Anthropic to become the default AI provider for the U.S. government.

Meanwhile, Anthropic is fighting back by making it easier for millions of users to abandon ChatGPT and by building a new generation of autonomous AI workers.

In other words, this AI Cold War is now being fought on multiple fronts.

And you’re in the crossfire.

Washington Picks a Winner

The most visible battlefield is in Washington, where earlier this year a dispute between the Pentagon and Anthropic exploded into a full-blown political fight.

Anthropic’s CEO, Dario Amodei, refused to remove safety restrictions from the company’s Claude AI models. Those rules prohibit two things: using AI for domestic surveillance of Americans and deploying it in fully autonomous weapons.

Defense officials wanted those limits removed so the system could be used for “all lawful purposes.”

Amodei said no.

Within days, the Department of Defense escalated the conflict. Defense Secretary Pete Hegseth threatened to cancel Anthropic’s contracts and invoked federal supply-chain laws that are normally used against foreign adversaries.

Then the situation escalated again.

In early March, the Pentagon formally labeled Anthropic a “supply-chain risk.” The Trump administration then ordered federal agencies to phase out the company’s technology across government systems.

That decision effectively shut Anthropic out of a huge portion of the federal market.

And it created an opening for its biggest rival.

Almost immediately, OpenAI stepped in.

Through the General Services Administration’s OneGov purchasing program, federal agencies can now deploy ChatGPT Enterprise for about $1 per agency per year starting in 2026.

One dollar.

That symbolic pricing removes the biggest barrier to adopting AI inside the government. Agencies no longer need months of procurement negotiations to test the technology.

They can simply turn it on.

OpenAI followed that move with another major win.

The company secured a Department of Defense contract worth up to $200 million through its “OpenAI for Government” program, placing its models directly inside national-security workflows.

Put those developments together and CEO Sam Altman’s strategy becomes clear.

OpenAI isn’t just selling software.

It’s positioning ChatGPT as the default AI platform for federal work.

History shows how powerful that position can be. Once a technology platform gets embedded across government systems, replacing it becomes extremely difficult. Microsoft Windows dominated federal computers for decades for exactly that reason.

But Washington isn’t the only battlefield where this rivalry is playing out.

Earlier this year, Anthropic aired its first-ever Super Bowl commercials for Claude.

The ads took a clear shot at OpenAI’s plans to introduce advertising into ChatGPT. They implied that once chatbots depend on advertising, the advice they give could start sounding a lot like marketing.

OpenAI’s CEO Sam Altman fired back on X, calling the commercials “dishonest” and “deceptive.” He insisted OpenAI would never run ads in the way Anthropic portrayed them.

The exchange made it clear that this rivalry is no longer just about technology.

It’s about who controls the business model of AI.

And that fight is starting to spill over into user behavior.

App data recently showed Claude surging to #1 in Apple’s App Store, helped by a wave of “Cancel ChatGPT” posts after OpenAI’s defense contracts became public.

Even so, the U.S. military continued deepening its reliance on OpenAI’s models.

Which highlights how divided this market may become.

Inside the government, OpenAI is consolidating power.

Outside the government, Anthropic is trying to win something bigger.

OpenAI’s users.

Anthropic recently introduced tools that let users export their ChatGPT history and convert it into a “cognitive profile,” which summarizes their tone, preferences and workflows.

Claude can import that profile into its memory system.

Instead of training a new assistant from scratch, Claude can pick up where ChatGPT left off.

And Anthropic is pairing that strategy with something even more ambitious.

Autonomous AI agents.

Earlier this year, Anthropic introduced a desktop system called Claude Cowork that can actually operate software on your computer.

It can open files, browse websites, update documents and complete multi-step tasks.

Its capabilities run on something called the Model Context Protocol, or MCP, that allows AI models to connect directly to tools and data across different software platforms.

Developers are already integrating MCP into workplace apps like Slack, Figma and Asana so AI agents can interact with those systems.

Think of it less like a chatbot and more like a junior employee working alongside you.

Because that’s the real prize in this Cold War.

To be the platform that people — and eventually machines — rely on to get work done.

And right now, OpenAI and Anthropic appear to be taking two different paths toward that future.

Here’s My Take

Over the next few years, the AI industry could begin splitting into two very different ecosystems.

Governments and heavily regulated industries will likely standardize around a small number of approved AI platforms. Right now, OpenAI is moving quickly to secure that position in Washington with federal contracts and near-zero pricing.

But outside the government, a bigger battle will be fought over autonomous agents that can do work across software tools.

If Anthropic succeeds in turning Claude into the intelligence layer for those agents, a second ecosystem could emerge driven by developers, startups and everyday users.

This battle is far from over.

But whichever side becomes the platform regular people rely on to get work done will likely shape the next era of computing.

Regards,

Ian King
Chief Strategist, Banyan Hill Publishing

Editor’s Note: We’d love to hear from you!

If you want to share your thoughts or suggestions about the Daily Disruptor, or if there are any specific topics you’d like us to cover, just send an email to [email protected].

Don’t worry, we won’t reveal your full name in the event we publish a response. So feel free to comment away!