The AI boom is starting to fade, and financial pressure is forcing many Silicon Valley star startups to choose to exit.

Key points:

1. Several AI startups that have raised billions of dollars in funding have begun to scale back their operations, attempting to compete with industry giants is like hitting a rock with an egg.

2. In the past three years, investors have poured $330 billion into approximately 26,000 AI and machine learning startups.

3. OpenAI is the most successful company among AI startups, with revenue reaching $16 billion last year.

4. Once thriving Stability AI has restructured and downsized its operations, while Inflection AI has closed its business and been largely integrated into Microsoft.

Tencent Technology News, May 4th - Foreign media recently pointed out in an article that the AI boom has gradually begun to fade. The threshold for startups to compete with tech giants such as Microsoft and Google in the field of AI has risen to billions of dollars. Even with such a large amount of funding, it may still be insufficient to compete with these tech giants.

Since mid-March this year, several once thriving Silicon Valley AI startups have suffered heavy financial pressure. Inflection AI, which raised $1.5 billion but almost generated no revenue, has abandoned its original business. Stability AI, which gained great popularity with the launch of the open-source large model platform Stable Diffusion, has not only laid off employees but also dismissed its CEO. Anthropic, which received huge investments from Amazon and Google, has been struggling to narrow the gap of nearly $1.8 billion between its revenue and massive expenses.

It is becoming increasingly clear in Silicon Valley that the AI revolution will come at a huge cost. Tech companies betting on this revolution are striving to find ways to narrow the gap between massive expenses and future profits. For some star startups that have collectively raised hundreds of billions of dollars for developing general AI, this issue is particularly serious. Some startups have realized that competing with giants like Google, Microsoft, and Meta may require billions of dollars in expenses. Even so, such a scale of funding may still be insufficient.

"You can already see the signs of trouble," said Ali Ghodsi, CEO of Databricks, a data warehouse and analytics company that collaborates with AI startups. "It's not just about how cool what you're doing is, but whether it's commercially viable."

Although other tech trends have also burned through a lot of money, the cost of building AI systems still shocks industry veterans. Unlike the iPhone, which ushered in the last technological transformation, the cost of creating and maintaining generative AI models can reach billions of dollars. In contrast, the transformation sparked by the iPhone largely relied on existing components and could be developed with just hundreds of millions of dollars. The advanced chips needed by AI startups are expensive and in short supply, and each query of an AI system costs more than a simple Google search.

According to data provided by the database analysis platform Pitchbook, investors have poured $330 billion into approximately 26,000 AI and machine learning startups over the past three years, two-thirds more than the amount they funded for 20,350 AI startups from 2018 to 2020.

Many emerging AI companies are currently facing challenges that sharply contrast with the early business achievements of OpenAI, which received $13 billion in support from Microsoft. The huge success of ChatGPT has enabled OpenAI to charge $20 per month for this high-end chatbot service. Insiders revealed that OpenAI's revenue reached about $1.6 billion last year, but the company's expenditure data is still unknown. OpenAI has not commented on this report so far.

Even startups like OpenAI are facing challenges in expanding their revenue. Enterprise users of the company are concerned that AI systems may produce inaccurate answers. In addition, OpenAI is also troubled by whether the data supporting its models infringes copyrights.

Many investors have pointed out that Microsoft's rapid revenue growth demonstrates the commercial potential of AI. Brad Reback, an analyst at investment bank Stifel, stated that in the most recent quarter, Microsoft's estimated revenue from cloud computing AI services was $1 billion, compared to almost zero a year ago.

On the other hand, although Meta has increased its infrastructure spending by as much as $10 billion this year, the company does not expect to make money from its AI products for several years. Meta CEO Mark Zuckerberg said in a recent earnings call with analysts, "We are investing to maintain a leading edge in the field of AI."

AI startups have long been troubled by the significant gap between expenses and revenue. Two sources familiar with Anthropic's financial situation revealed that the company has raised over $7 billion with the support of Amazon and Google. Although the company's current annual revenue has reached $150 million to $200 million, its annual expenditure is as high as $2 billion.

Like OpenAI, Anthropic has turned to collaborate with large established tech companies. The company's CEO, Dario Amodei, has been courting Wall Street clients. Recently, the company announced a partnership with global consulting firm Accenture to develop custom chatbots and AI systems for enterprises and government agencies. Sally Aldous, a spokesperson for Anthropic, stated that thousands of companies are using the company's technology, and millions of consumers are using the chatbot Claude developed by the company.

Stability AI, a startup specializing in image-generating AI, announced last month that its co-founder and CEO, Emad Mostaque, had resigned from the company. Just a week ago, three researchers from the five-person team that developed the company's original technology chose to resign. Insiders familiar with Stability AI's business revealed that the company is expected to achieve revenue of about $60 million this year, while the cost of its image-generating system is about $96 million.

Artificial intelligence investors say that the financial situation of Stability AI looks better than that of language model developers like Anthropic, because the cost of developing image generation systems is lower. However, at the same time, with the gradual decrease in demand for purchasing images, the sales prospects of Stability AI are becoming more uncertain.

Stability AI has been operating without the support of tech giants. Two former employees of the company stated that after raising $101 million from venture capitalists in 2022, Stability AI needed more funding last fall, but had difficulty proving to investors that it could sell its technology to businesses. Ultimately, Stability AI received a $50 million investment from Intel at the end of last year. Even so, the company still faces significant financial pressure.

The report also stated that as Stability AI's business develops, its sales strategy has changed. At the same time, its monthly investment in computing power reaches millions of dollars. In this situation, some investors pressured Mostaq to resign. After Mostaq resigned, Stability AI carried out layoffs and business restructuring to put the company on "a more sustainable path." As of now, Mostaq and Stability AI have not commented on this report.

05 Inflection AI basically merged into Microsoft

Inflection AI, a chatbot startup founded by three senior professionals in the artificial intelligence industry, has raised $1.5 billion from some big names and well-known companies in the tech industry. However, insiders revealed that a year after launching the AI personal assistant, Inflection AI still had almost no revenue. In a letter to investors, Inflection AI stated, "Given the current AI market bubble, further financing is not the best option." At the end of March this year, Inflection AI closed its original business and basically merged into Microsoft.

Microsoft had also participated in the financing of Inflection AI. Mustafa Suleyman, CEO of Inflection AI, is one of the founders of DeepMind, a pioneering AI lab acquired by Google in 2014. Suleyman, along with Karen Simonyan, an important researcher at DeepMind, and Reid Hoffman, a well-known venture capitalist in Silicon Valley (who helped establish OpenAI and is now a member of Microsoft's board), co-founded Inflection AI. As of now, Inflection AI and Microsoft have not commented on this report.

Inflection AI had many talented AI researchers who had previously worked at companies such as Google and OpenAI. But after releasing the AI personal assistant for nearly a year, as one investor put it, "Inflection AI's revenue is 'almost negligible,' basically zero. Unless it continues to raise huge amounts of funding, it cannot continue to improve its technology and compete with upgraded products from companies like Google and OpenAI."

Now, Microsoft has absorbed most of the Inflection AI employees, including Suleyman and Simonyan. This cost Microsoft over $650 million. However, unlike Inflection AI, Microsoft has the ability to fight a long battle. The company has announced plans to establish an AI lab in London for employees to collaborate with startups hoping to make technological breakthroughs. (Compiled/Unconstrained)