There’s an AI Price War Brewing Between Anthropic and OpenAI – Why That Could Burst the Bubble

OpenAI is considering lowering the price of tokens used to power its artificial intelligence (“AI”) models, such as ChatGPT, according to a report from the Wall Street Journal. Worse, key rival Anthropic also expects to lower its price. A price war could upend the burgeoning AI industry, where investors are already worried about profitability.

Even if an all-out price war doesn’t come to pass, the news doesn’t look so hot for AI either way.

The news comes as OpenAI has reportedly missed its own sales goals and is failing to grow customer counts fast enough. OpenAI is worried about meeting its own huge spending commitments if sales can’t rise faster. OpenAI has locked in $1.4 trillion in spending, per Barron’s, including deals with Microsoft (MSFT) and Oracle (ORCL), among others.

In recent weeks, there has been growing speculation about a government bailout or intervention for OpenAI. That’s just not the kind of news that healthy companies would want popping up in the press.

As for an AI price war, it couldn’t happen at a worse time for either company. Both have raised startling amounts of cash – OpenAI $122 billion in a March round and Anthropic $65 billion in a May round. Those fundraising rounds pushed the firms’ valuations to stratospheric highs, with OpenAI valued at $852 billion and Anthropic to just shy of $1 trillion, at $965 billion.

But with the utterly massive investment demands on both companies and incredible cash-burn rates, they can ill afford to get thrust into a price war. These companies have already been heavily subsidizing users, and undercharging them further would hurt both rivals.

Anthropic recently revealed that it expects to earn $559 million in the second quarter when it filed the prospectus for its initial public offering (“IPO”) with the Securities and Exchange Commission. Cynics might see a connection between the two. It quickly walked back talk of being profitable for the rest of the year, however, as it spends aggressively on growth.

OpenAI’s profitability looks worse. The firm generated sales of $5.7 billion in the first quarter, but had an operating profit margin of negative 122%, according to The Information. So, for each dollar of sales, OpenAI lost $1.22. OpenAI won’t “make it up on volume” by stealing users from its rival at an even lower cost than it’s already getting.

Of course, Anthropic’s spotty profitability would also suffer if a price war breaks out.

Even if it doesn’t, the potential for two key AI players to consider a price war suggests that the industry is in much worse shape than the soaring valuations and their IPO announcements would indicate.

Looming AI Price War Threatens to Sink the Tech Industry

For companies to even consider a price war suggests something profoundly negative is going on in the industry. Price wars are unhelpful to almost everyone, and rational economic players don’t launch them unless they think they have no other good alternative – and they may not.

A price war when companies are already having severe profitability issues is a kind of “grasping at straws to keep from drowning” moment.

Even OpenAI CEO Sam Altman himself said the cost of AI had become “a huge issue.”

Spending commitments at both companies mean that they need to keep sales growing briskly. So, they’re feeling the pressure to push sales higher at any price, even their own profitability. But if both companies are feeling the heat, it suggests the industry as a whole may not be growing as fast as predictions indicate. Rivals are rushing to get market share while they can.

None of this is the kind of dynamic that you expect to see in an industry that has been as highly touted as AI, at least not this early in its life cycle.

This kind of irrational economic behavior can be sustained only with a pile of investors’ capital that’s willing to subsidize this growth. When that dries up, AI prices have to go up so the price of service can sustainably cover their costs, or the industry will need to “consolidate” – that is, supply will need to come out of the market so prices can rise (i.e., companies may fail).

Alternatives Are Pressuring Anthropic and OpenAI for Paying AI Customers

Here’s the thing: AI companies are likely to face even greater pricing pressure in the future. Many AI tasks don’t need the pricey, high-end “frontier” models that cost the most to run. Low-end AI models work for many tasks, particularly for everyday users with routine requests.

Worse, AI companies may soon discover that they can’t even get everyday users to pay up for service. Consumers have access to thousands of free AI models for download, and laptops can run them. So, consumers may not be motivated to pay for much more than an AI software app subscription – let alone a per-use fee – as AI apps go “on-device” at little or no incremental cost.

We’re already seeing Alphabet (GOOGL) push AI onto devices, a move that begins to undercut users who turn to ChatGPT or Claude for garden-variety queries. Of course, Google Search also uses the top portion of its real estate to deliver AI-generated summaries, before users go to an AI tool.

While Alphabet is using its positional advantage to bring AI closer to users, it’s also using its robust financial strength to subsidize AI in ways that OpenAI and Anthropic can’t. If AI is your only product, as it is for AI startups, this product must succeed for the company to succeed. But Alphabet can deliver AI, even at a loss, as one part of an already profitable suite of products.

Alphabet’s massive $80 billion equity raise shows it’s going “pedal to the metal” on AI spending.

So, OpenAI and Anthropic have leaned hard into the enterprise business, where they may be able to lock in users who are willing to pay significant amounts for higher-end models.

The situation there is looking less attractive, however. More reports are emerging in the press about companies discovering that AI is not proving effective at driving productivity or other success metrics. Some companies have pulled back on so-called “token-maxxing,” or the practice of using AI as much as you want.

The return on investment may simply not be there. Ride-hailing firm Uber (UBER) said it has not been able to link its increased use of AI to the development of customer-serving innovations.

Meanwhile, Meta Platforms (META) is set to spend billions in AI for internal use alone just this year. An internal memo indicated that it’s set to curb and manage AI spending in 2027, after previously encouraging employees to token-maxx, according to The Information.

Add it all up, and it doesn’t look good for the AI model companies.

Consumers have a viable and cheaper alternative. Major rivals have positional and financial advantages. Enterprise customers are beginning to look closely at the effectiveness of their AI spending. That’s not a recipe for AI models turning profitable on a sustainable basis, but it sure helps explain why they may be looking to increase market share while they can.

It also helps explain why major AI players, including SpaceX (SPCX), Anthropic, and OpenAI, have all recently filed for an IPO while investors’ animal spirits are running hot for AI. Even as valuations are rising, the economics may be poor and deteriorating – so venture capital investors may think it’s a good time to sell to the public.

Regards,

James Royal, PhD

Editor’s Note: Every generation or so, the way money moves gets a fundamental upgrade. The people who see it coming have the chance to get extraordinarily wealthy. Everyone else watches from the sidelines. Luke Lango says that moment is here again, and Elon Musk is behind the upgrade in an amazing way. He’s revealing exactly what to buy, including one free pick, in this presentation.