OpenAI Crisis: The Full Story of Money, Users, and Rivals

01What the OpenAI crisis is and how it started

The OpenAI crisis is a triple problem: massive cash burn, a user revolt after a Pentagon deal, and rising competition from rivals like Claude. These three issues are connected and happened at the same time.

The crisis began with money. OpenAI expects to burn through $218 billion by 2029. To raise cash, OpenAI signed a deal with the Pentagon, which triggered a user backlash.

That backlash boosted Anthropic's Claude app to the top of the App Store. This chain reaction shows how financial pressure can lead to strategic mistakes and user distrust.

In this episode, Julian Goldie breaks down the numbers and explains what it means for anyone using AI tools in business. Understanding the full story helps you make better decisions about which AI tools to rely on.

02The money problem: $218 billion in expected losses

OpenAI generated about $13.1 billion in revenue in 2025. But in the first half of that year, it posted a net loss of $13.5 billion. That means it lost more money in six months than it made in the entire year.

OpenAI spends $1.69 for every dollar of revenue. By 2026, cash burn is expected to reach $25 billion, and $57 billion in 2027. Cumulative burn through 2030 is projected at $665 billion, with no positive cash flow expected until then.

Unlike Amazon or Tesla, OpenAI builds no physical assets. Its product evaporates, and users can switch to a rival in 30 seconds. There's no network effect to protect them.

03Why AI training and hardware costs explode

AI does not scale like normal software. Training a model can cost $100 million for one run, but current models can cost over $1 billion per training run. If a competitor releases something better, OpenAI must train again, and each time the bill gets bigger.

Hardware adds to the cost. Each NVIDIA Blackwell B200 chip costs $30,000 to $40,000, and you need tens of thousands. These chips have a shelf life measured in months, not years.

Project Stargate, a $500 billion supercomputer project, requires the power of multiple nuclear reactors.

Every free ChatGPT user increases the cost. Hundreds of millions of people on the free tier cost OpenAI real money every session. The product works, but the bills are impossible to sustain.

04Microsoft and NVIDIA: money goes in circles

Microsoft invests billions in OpenAI, but a large part never leaves Microsoft. Instead of cash, Microsoft gives OpenAI cloud credits for Azure. OpenAI records it as capital raised, but those credits can only be spent at Azure, recycling the investment back to Microsoft.

This creates a problem: you can't pay employees with gift cards. OpenAI needs actual dollars for salaries, so it has to raise hard cash from outside investors every quarter.

NVIDIA also invests in OpenAI, up to $100 billion, but that money goes right back to NVIDIA for GPU purchases. NVIDIA also invests in CoreWeave, which supplies cloud capacity to OpenAI. The cycle is circular, and the real question is whether OpenAI can break out before the loop breaks them.

05The three-front war: Meta, China, and user choices

OpenAI faces three competitors that make it hard to charge for AI. First, Meta releases its Llama models for free. Zuckerberg doesn't need AI to be profitable—he just sets a ceiling on what OpenAI can charge.

Second, Chinese models cost as little as one-sixth of comparable US systems. DeepSeek's V4, a one-trillion-parameter coding model, launched in February 2026 at a fraction of the cost. That breaks OpenAI's assumption that you need to spend tens of billions to stay at the frontier.

Third, users are switching. OpenAI's web traffic share fell from 86% in January 2025 to 64% in January 2026. Gemini grew from 5% to 21%.

About 75% of OpenAI's revenue comes from consumer subscriptions, but cancellations are rising. Only 20-30% of enterprise users stick with their API long-term. Users have no lock-in.

06The Pentagon deal backfired: 295% uninstall surge

On February 27, 2026, the Department of Defense asked both OpenAI and Anthropic for unrestricted military access. Anthropic's CEO Dario Amodei said no publicly, stating some uses are outside what today's technology can safely do.

Hours later, Sam Altman signed the deal. The internet exploded. The QuitGPT boycott gained over 2.5 million pledges.

By March 1, ChatGPT uninstalls jumped 295%, and one-star reviews surged 775% in a single day.

OpenAI reportedly lost about 1.5 million paid subscribers in the first week. At $20 per month, that's $30 million in lost monthly recurring revenue. Downloads of Claude jumped 51% in one day, and by Sunday it hit number one on the US App Store—the first time ever.

07Claude's rise: one million new signups a day

Anthropic confirmed daily active users have more than tripled since the start of 2026, and paid subscribers doubled. That means over one million new signups per day.

ChatGPT still has over 300 million weekly active users, so one boycott week doesn't erase its base. But it shows loyalty is gone. Users are not just switching off protests; they are switching and staying.

The first time Claude is good enough, the switch becomes permanent. That's a different world than what OpenAI built its business plan around. Anthropic gained by declining a $200 million government contract, showing that users care about how AI tools are used.

08The AGI bet and the risk of bankruptcy

Underneath the losses and controversy is a single bet: Artificial General Intelligence (AGI). If OpenAI builds an AI that can do any intellectual task a human can do, the cash burn won't matter. Revenue would be unlimited.

But the math is brutal. Losing $14-17 billion a year means every month of delay costs over a billion dollars. If AGI takes five years instead of two, the funding gap could reach nearly $100 billion.

HSBC analysis says OpenAI likely won't make money by 2030 and will face a $207 billion shortfall. Cash reserves could run out by mid-2027. The likely outcome isn't a crash but a quiet sale, probably to Microsoft, which has over $80 billion in cash reserves.

09Talent war: engineers leave and retention costs soar

Engineers at OpenAI are eight times more likely to leave for Anthropic than the reverse. Senior executives like CTO Mira Murati and co-founder Ilya Sutskever are among those departing, many to start rivals or join Anthropic.

To keep employees, OpenAI expects to spend $6 billion on stock-based compensation, nearly half of projected revenue. These packages only work if the valuation keeps climbing.

If investors lose faith, those stock packages lose value, and researchers will do the math and head for the door. The talent exodus makes it even harder to achieve the AGI breakthrough OpenAI needs to survive.

10What the OpenAI crisis means for you

If you use AI tools in your business, this story matters. OpenAI is not finished—it has hundreds of millions of users, billions in the bank, and the most recognizable brand in AI. But the landscape is changing.

Users are more informed and loyalties are shifting. Anthropic gained by saying no to a military contract and won with a million new signups a day. That shows the market rewards ethical behavior.

Stay ahead by learning the tools and understanding the landscape now. By the time it's obvious what happened, it will be too late. Pay attention, and make informed choices about which AI tools you trust.