You could say it's the end of the beginning of the AI boom.
Since mid-March, financial pressures on some leading artificial intelligence startups have taken a heavy toll. Inflection AI raised $1.5 billion but made little profit and folded its original business. Stability AI has laid off employees and parted ways with its CEO. And Anthropic is scrambling to close the roughly $1.8 billion gap between modest sales and massive costs.
Silicon Valley is making it clear that the AI revolution will come at a very high cost. And the tech companies that are betting their future on it are scrambling to figure out how to close the gap between those costs and the future profits they hope to make.
The problem is particularly acute for a group of high-profile startups that have raised tens of billions of dollars to develop generative AI, the technology behind chatbots such as ChatGPT. Some of them already understand that it will take billions of dollars to compete head-on with giants like Google, Microsoft, and Meta, but even that may not be enough.
“We can already see the writing on the wall,” said Ali Ghodshi, CEO of Databricks, a data warehousing and analytics company that works with AI startups. “It doesn’t matter how cool what you do, is it viable as a business?”
While large sums of money have been spent in other technology booms, the cost of building AI systems is shocking tech industry veterans. Unlike the iPhone, which was the beginning of the last technology transition and cost hundreds of millions of dollars to develop because it relied heavily on existing components, generative AI models cost billions of dollars to create and maintain. . The cutting-edge chips they need are expensive and in short supply. And every query for an AI system costs much more than a simple Google search.
Investors have poured $330 billion into about 26,000 AI and machine learning startups over the past three years, according to industry tracker PitchBook. This is more than two-thirds of the amount spent funding 20,350 AI companies between 2018 and 2020.
The challenges facing many emerging AI companies stand in contrast to the early achievements of OpenAI, which has received $13 billion in backing from Microsoft. The attention generated by the ChatGPT system has allowed the company to build a business that charges $20 per month for its premium chatbots, and uses the technology that powers its chatbots, called Large Language, to build AI services. We provided companies with a method. model. OpenAI generated about $1.6 billion in revenue last year, but it's unclear how much the company is spending, two people familiar with the company's business said.
OpenAI did not respond to a request for comment.
However, even OpenAI faced challenges in increasing sales. Companies are wary that AI systems can generate inaccurate answers. The technology has also been plagued by questions about whether the data supporting the models infringes copyright.
(The New York Times sued OpenAI and Microsoft in December for copyright infringement of news content related to AI systems.)
Many investors point to Microsoft's rapid revenue growth as evidence of AI's business potential. Microsoft reported an estimated $1 billion in revenue for AI services in cloud computing in its most recent quarter, compared to virtually zero a year ago, said Brad Riback, an analyst at investment bank Stifel. However, this number has increased.
Meta, on the other hand, doesn't think its AI products will make money for years, even if it increases infrastructure spending by up to $10 billion this year alone. “We're investing to stay on the cutting edge of this space,” Meta CEO Mark Zuckerberg said on a conference call with analysts last week. “And we do that when we scale a product before it becomes profitable.”
AI startups are suffering from a gap between spending and revenue. Anthropic, which has raised more than $7 billion with backing from Amazon and Google, spends about $2 billion a year but generates only about $150 million to $200 million in revenue, the company said. Two people familiar with the finances spoke on condition of anonymity. The numbers are not disclosed.
Similar to OpenAI, Anthropic is looking to partner with large, established technology companies. The company's CEO, Dario Amodei, works with Wall Street clients and recently worked with Accenture, a global consulting firm, to develop custom chatbots and AI systems for businesses and government agencies. announced that it was doing so.
Anthropic spokeswoman Sally Aldous said thousands of businesses use the company's technology and millions of consumers use its publicly available chatbot, Claude. Ta.
Image generator Stability AI lost its founding chief executive Emad Mostaq last month, just one week after three researchers who were part of the five-person team that built the company's original technology resigned. announced.
The company expects to generate about $60 million in revenue this year, compared to the cost of about $96 million for the image generation systems it has been offering to customers since 2022, according to people familiar with the company's business.
AI investors say Stability AI's financial position looks better than language model makers like Anthropic because of lower development costs for its image generation systems. But the outlook for sales is more uncertain, as the demand to pay for images has also declined.
Stability AI has been operating without the help of tech giants. After raising $101 million from venture capitalists in 2022, they needed more money last fall, but were having trouble showing investors that their technology could be sold to companies, a former employee said. Two members spoke. They refused to speak publicly because they were not authorized to do so. Despite raising $50 million from Intel late last year, the company said it still faces financial pressures.
As the startup has grown, its sales strategy has changed, according to people familiar with the matter. At the same time, they were spending millions of dollars every month on computing costs. One investor said some investors pressured Mostak to resign, but he refused to discuss personnel matters publicly. After his departure, Stability AI cut jobs and restructured its business this month to put the company on a “more sustainable trajectory,” according to a company memo reviewed by The New York Times.
Stability AI declined to comment. Mostak declined to discuss his own departure.
Inflection AI, a chatbot startup founded by three AI veterans, has raised $1.5 billion from big tech companies. But one year after introducing the AI personal assistant, the company had almost no revenue, according to one investor. The Times reviewed a letter Inflexion sent to investors, saying that raising additional funding “is not the best use of investors' money, especially in the current frothy AI market conditions.” .
In late March, the company folded its original operations and was largely wiped out by Microsoft, the world's most valuable publicly traded company.
Microsoft also helped fund Inflection AI, whose CEO Mustafa Suleiman is the founder of DeepMind, a seminal artificial intelligence research lab acquired by Google in 2014. He became famous as one of the. Suleiman founded his Inflection AI with Karen Simonyan, a principal researcher at DeepMind. , Reid Hoffman is a leading venture capitalist in Silicon Valley, helped found OpenAI, and serves on the board of Microsoft.
Microsoft and Inflection AI declined to comment.
The company had many talented AI researchers who had worked at Google, OpenAI, and others.
But nearly a year after releasing its AI personal assistant, Inflection AI's revenue was, in the words of one investor, “minimal.” Basically zilch. If we don't continue to raise huge amounts of funding, we won't be able to continue to improve our technology and catch up with chatbots like Google and OpenAI.
Microsoft has now swallowed most of its employees, including Mr. Suleiman and Dr. Simonian.
This will cost Microsoft more than $650 million. But unlike Inflection AI, it has the luxury of being long-lasting. The company announced plans for staff to build an AI lab in London to work on the kinds of systems startups hope to make breakthroughs with.
erin griffith Contributed to the report.
