Investors Are Flooding Into AI Startups Despite Lack Of Profits - or Revenues

Among venture investors there appears to be a bet that demand for AI capabilities of whatever focus is so great and so assured that no enterprise will go unrewarded either by acquisition, merger or initial public offering. 

The result has been unprecedented inflows of money into AI, Gen AI and anything else related. While not necessarily a sure thing or even a safe bet, it appears to feel to many like as good an investment promise of excess returns as is currently available. We've been here before and history teaches that there are always losers as well as winners, but accumulations of wealth of late have been so great that many feel insulated from the consequences of their choices right now. JL 

Berber Jin reports in the Wall Street Journal:

Despite a broad downturn in the startup sector, investors have deluged AI upstarts with record levels of funding, minting dozens of companies with billion-dollar valuations in the past year. VCs are betting some startups will be the pioneers in a tech revolution that could outshine the birth of the Internet. (But) few startups have been able to replicate OpenAI's success. The investment frenzy is fueling concerns of a bubble as startups struggle to translate the hype into revenue. Last year, investors poured $21.8 billion into generative AI deals, up fivefold from the prior year. The average round size for those deals was $51 million, compared with the industry average of $8 million. “The problem is that we don’t know what these business models are going to look like at scale."

Artificial intelligence startup Imbue has hoodies branded with its circular orange logo, an office in the heart of San Francisco and marquee investors who lavished the company with more than $210 million.

Work and life blend together for its few dozen employees, who share their emotions with one another at a weekly event called “Feelings Friday” to build trust and connection. 

More than two years into its founding, what the startup doesn’t have is a business—or a product that could create one.

 

Despite a broad downturn in the startup sector, investors chasing the stock market successes of Nvidia and Microsoft have deluged AI upstarts with record levels of funding, minting dozens of companies with billion-dollar valuations in the past year. The investment frenzy is already fueling concerns of a bubble as startups struggle to translate the hype into revenue.

“Everyone believes that AI is the future, so we are going to see an extraordinary amount of investment until proven otherwise,” said Alex Clayton, a general partner at the venture firm Meritech. “The problem is that we don’t know what these business models are going to look like at scale. You can have theories about it, but you really don’t know.”

Fears of rising startup valuations aren’t new in Silicon Valley. But the AI gold rush is notable because investors are writing massive checks—sometimes in the hundreds of millions of dollars—just to get these companies off the ground. Even during the peak of the startup boom, such large financings were reserved for later-stage private companies gearing up for aggressive growth.

Imbue hit a valuation of more than $1 billion with its fundraising last year, courting backers like Nvidia and ex-Google CEO Eric Schmidt. Chief Executive Kanjun Qiu dazzled investors with a vision to build intelligent computers that could give humans the “freedom, dignity, and agency to do the things we love.” Last November, she lavished her employees with a company off-site to Japan.

Imbue is plowing its cash into developing AI models that it hopes will one day create autonomous AI agents. A company spokesperson said it made a “deliberate strategic decision” not to commercialize in order to focus on research and that investors were on board with this approach. 

Unlike traditional software companies, the language models underpinning generative AI apps like conversational chatbots are expensive to build. Training them requires data centers with electricity needs so intense that industry leaders are warning that power grids won’t be able to keep up with the demand.

Last year, investors poured $21.8 billion into generative AI deals, up fivefold from the prior year, according to the research firm CB Insights. The average round size for those deals was $51 million, compared with the industry average of $8 million. Tech giants including Microsoft and Amazon contributed to the upswell in funding—with the added benefit of seeing their investments flow back to them through cloud-computing contracts.

Venture capitalists are betting that some of these startups will be the pioneers in a tech revolution that could outshine even the birth of the Internet. They point to the meteoric rise of OpenAI, whose chatbot ChatGPT became the fastest-growing consumer app in Internet history. OpenAI went from zero to more than $1 billion in revenue last year, a brisk growth rate even by the breakneck standards of Silicon Valley. 

So far, few other startups with similarly large ambitions have been able to replicate that success.

One example is Inflection AI, the buzzy AI startup co-founded by LinkedIn founder Reid Hoffman and backed by tech billionaires, including Bill Gates. In the year after it was founded, Inflection raised $1.5 billion to develop the language models powering its main product, a chatbot called Pi that gave emotional support to its users. 

But the company couldn’t find a business model that worked. Last month, Inflection Chief Executive Mustafa Suleyman and most of his staff decamped to Microsoft, leaving Inflection a shell of its former self. The startup has a new CEO and is trying a new strategy of selling its software to businesses. 

Other startups that raised funding rounds of more than $100 million generate little or no revenue. They include the digital avatar creator Character AI, which raised $150 million last year from investors including the venture firm Andreessen Horowitz; and Magic AI, a coding assistant startup that raised $117 million in February. 

The Inflection website as seen on a smartphone. PHOTO: GABBY JONES/BLOOMBERG NEWS

Some generative AI startups that received early funding from venture capitalists, such as Jasper and Tome, have also had to lay off staff after seeing slower-than-expected revenue growth. 

“It’s almost a curse if you’re an early AI company and you have revenue, because then you get valued on the numbers, not just the story,” said Clayton of Meritech Capital.

The funding rush, paired with a lag in product adoption, has resulted in skewed economics for generative AI startups.

At its annual AI summit in March, the venture firm Sequoia Capital estimated that the industry sank around $50 billion into the state-of-the-art Nvidia chips needed to train language models. By contrast, generative AI startups brought in $3 billion in revenue.

“Everybody’s assuming: if you build it, they will come. AI is a field of dreams,” said Sonya Huang, a Sequoia partner, at the event. “The amount of money it takes to build this stuff has vastly exceeded the amount of money coming out so far. So we’ve got some real problems to fix.”

Venture capitalists including Huang predict that revenue will take off as the technology becomes cheaper and more reliable. There are also early signs that businesses are beginning to embrace generative AI: in February, the buy-now-pay-later app Klarna said that an AI assistant powered by OpenAI was doing the same work as 700 customer service agents.

But most companies have been hesitant to increase spending on generative AI, which remains error-prone and expensive to use. The limitations have led some investors to urge a more cautious approach to valuing startups selling such services.

“For the companies that raised 2, 3 years back, you’re seeing that revenues are not catching up with valuations,” said Umesh Padval, a managing director at the venture firm Thomvest Ventures. “The use cases are still not there.”

Brad Gerstner, founder of investment firm Altimeter Capital, which has backed tech hits like Uber and Snowflake, said in a March podcast that generative AI revenue should be viewed as experimental rather than the stable recurring revenue typically used to value software companies.

At Altimeter, when discussing revenue from AI companies, investors are banned from using the term ARR, which stands for annual recurring revenue and typically describes money made from software subscriptions, Gerstner said. 

Not everyone is taking such a sober approach.Last fall, a team of young startup researchers working out of an apartment in New York’s Upper East Side began developing an AI tool that helps software engineers complete difficult coding tasks. 

They raised $21 million from investors including venture firm Founders Fund in March, just a few weeks after formally incorporating the business, called Cognition AI. The startup has lower research costs than others because it doesn’t build its own language models, instead relying on those provided by OpenAI. But it hasn’t publicly released a product. 

Cognition recently closed a new round of funding valuing it at $2 billion.