Every start-up is an AI company now. Bubble fears are growing. – The Washington Post

SAN FRANCISCO — Earlier this year, investment bank UBS put out a report heralding artificial intelligence start-up OpenAI’s ChatGPT chatbot, saying it was on pace to hit 100 million monthly users just three months after being released publicly.
That would beat social media app TikTok, which took nine months, and Instagram’s 2½ years to hit the same milestone, the analysts wrote. “In twenty years following the internet space, we cannot recall a faster ramp in a consumer internet app.”
While the report kicked off a wave of news coverage and excitement — a relief for a beleaguered tech industry struggling with a downturn — the 100 million number was based on website visits, not official “monthly active user” numbers from OpenAI itself, which would have been comparable to the TikTok and Instagram stats.
That report helped spark the AI fever that has gripped Silicon Valley for nearly a year now, triggering a gold rush as companies such as Google and Microsoft raced to compete and launch their own chatbots. Venture capitalists have poured billions of dollars into AI start-ups on spec, while in earnings calls over the past two weeks, tech CEOs have cited AI as a key growth area.
But it’s still unclear how and when this technology will actually become profitable — or if it ever will. There are already some reports that ChatGPT usage is declining. And “generative AI” is incredibly expensive to build and run — from specialized chips to data server computing power to expensive engineers.
“Everyone and their mom” is adding some kind of AI technology to their start-up to attract attention from venture capitalists, said Alice Deng, co-founder of payments start-up Slope.
“I’m just worried that so much noise gets to the point where it pops this bubble and then suddenly no one cares anymore,” Deng said.
Silicon Valley has seen wave after wave of hype cycle over the past decades — with mixed success. The infamous dot-com bubble saw companies go public and take hundreds of millions of dollars from retail investors simply for having a “.com” in their name. Countless social media companies battled for supremacy, but few remember names like FriendFeed and Yik Yak.
Meanwhile, billions of dollars have gone into building self-driving cars, but after years of development, the tech still isn’t ready for mass adoption, despite the predictions from tech luminaries that they would be widespread by the mid-2020s. Cryptocurrencies have seen their own cycles of excitement, most recently the bubble that peaked at the end of 2021, leading to millions of people losing money.
There was even a previous wave of AI venture capital hubbub in the mid-2010s, when scientific breakthroughs in image recognition, translation and other AI developments led to a burst of start-ups in the space. Many were acquired by Big Tech companies.
In the latest iteration, the tech behind chatbots such as ChatGPT and Google’s Bard are trained on huge amounts of data pulled from the open internet. That means big energy and computing needs. Nvidia, the company that makes the computer chips and software best suited to AI, has seen its valuation balloon over the last year, catapulting it into becoming the world’s sixth-most valuable company at $1.1 trillion.
Chatbots also make up false information and pass it off as real, a problem that the companies are trying to solve but some AI researchers say might not be possible to fix. AI companies have also come under fire from artists, filmmakers and musicians for using their copyrighted works to train AI models without permission or payment. A growing number of lawsuits seeks to stop the companies from using data from the open web, a key part of why the bots work so well.
Previous waves of consumer-focused tech like social media or e-commerce were enabled by relatively cheap access to online advertising and cloud storage. AI is more costly, making it harder for companies to succeed, especially if they haven’t figured out their business models yet.
“At the end of the day, AI is just software, it’s expensive software,” Andrew Harrison, CEO of venture capital firm Section 32, said. “It’s low-margin software unless it does something that’s 10 times better.”
The UBS report, which properly cited its data, generated buzz in part because the nuance was lost on people. Charts comparing ChatGPT to multibillion-dollar apps such as Instagram, Spotify, Uber and TikTok ricocheted across social media. Reporters started including the 100 million number in stories as fact, without explaining where it originally came from. Speakers reeled it off at conferences as evidence that AI would be as disruptive to society as the advent of the internet or electricity.
OpenAI does not disclose how many people actually use ChatGPT. A spokesperson for OpenAI declined to comment on the company’s user numbers.
Google hasn’t disclosed numbers for Bard either, or for how many people are using the version of its search engine that answers queries with AI-generated answers. Around 750,000 users of its productivity tools have been granted access to generative AI tools, but the company hasn’t said how many are actively using them.
In July, Microsoft announced pricing for some of its new generative AI products, including a tool that helps cybersecurity professionals understand what kinds of hacks their company may be facing. But it also hasn’t said how many people are using the tools yet.
Spokespeople for Google and Microsoft did not return requests for comment.
Over the past six months, AI has helped boost the valuations of Big Tech companies after a year of bad news, including layoffs and declining ad revenue. But while tech CEOs repeatedly cited the value of generative AI to their businesses during earnings calls over the past two weeks, there are still few details on how they will turn it into its own multibillion-dollar financial success.
In a call with analysts last week, Google CEO Sundar Pichai said the number of the company’s cloud customers who have signed up to use generative AI models available on its platforms grew by 15 times from April to June, but didn’t provide hard numbers.
In 2018, Pichai said that AI would have a more “profound” impact on human society than the discovery of fire did. In interviews, conference talks and calls this year, he’s repeatedly said that generative AI is a huge business opportunity for the tech giant, and that he’s working on weaving it into all of the company’s product lines.
Microsoft, which signed a multibillion-dollar deal with OpenAI to use the start-up’s tech in its own products, says generative AI is already winning it new business. Over 11,000 companies have used the OpenAI tools provided by Microsoft’s cloud division.
“That’s nearly 100 new customers added every day this quarter,” CEO Satya Nadella said last week on a call with analysts. The number of businesses signing up to use the company’s GitHub Copilot AI coding tool doubled quarter over quarter, he added.
On Thursday, Amazon CEO Andy Jassy sounded a note of caution during a company earnings call. The company has largely banked so far on its cloud services provider, Amazon Web Services, for its forays into generative AI.
“We’re a few steps into a marathon in my opinion,” Jassy said. “I think it’s going to be transformative, and I think it’s going to transform virtually every customer experience that we know. But I think it’s really early. I think most companies are still figuring out how they want to approach it.”
(Amazon founder Jeff Bezos owns The Washington Post. Interim CEO Patty Stonesifer sits on Amazon’s board.)
Apple CEO Tim Cook said in a May conference call that “there are a number of issues that need to be sorted” when it comes to the tech.
But on a Thursday call to discuss the company’s earnings with analysts, he elaborated, saying that the company had actually been working on generative AI “for years” and that AI was “integral to virtually every product that we build.”