Why Generative AI Startups Are So Dependent on Big Tech

BY SAM BLUM, SENIOR WRITER@SAMMBLUM

Generative AI has developed at a breakneck pace, and Silicon Valley's corporate giants do not want the technology to run away without them. 

In January, Microsoft's $10 billion investment in OpenAI catalyzed the funding stampede of 2023, ultimately setting the tone for generative AI's breakout year. The technology became an overnight sensation for its ability to absorb massive data sets and generate images and text via written user prompts. Despite a slowdown in venture financing generally, sparked by interest rate hikes and the bloated startup valuations of 2021, generative AI was a lightning rod of hype for investors this year. Globally, the sector netted $17.8 billion in 2023, according to Dealroom. 

Other giants have thrown billions at startups in a bid to carve out their places in the AI boom. Chipmaker Nvidia has participated in some of the sectors' biggest rounds, with investments in Inflection AI and Cohere, while Google's war chest actively pursued deals throughout 2023: The search giant took a piece of 16 startups this year, a Crunchbase analysis shows. In September, Amazon invested $4 billion in Anthropic, and retained a minority ownership stake in the company. Amazon's deal accounted for two-thirds of all generative A.I. funding in the U.S. during Q3, data from Pitchbook indicates. 

The need to partner with a bigger tech company has only grown as traditional venture funding in generative AI tapers off: The difficulties of startups Jasper and Stability AI to secure additional funding after large initial rounds triggered a kind of domino effect among investors, who are now more cagey about funding ambitious startups, the Information reported this month. 

Various experts have warned that the money flowing to AI startups from VCs would eventually run dry. But big tech has many incentives for its intensifying interest in generative AI's development, says Alejandro Lopez-Lira, a professor of finance at the University of Florida. It's a means of stymying competition from startups that may, conceivably, develop technology that could drive a breakthrough and leave them in the dust. Investment can obviously bring monetary reward, and the promise of generative AI is that transformational change--and a giant payday--could await savvy investors. 

"The fact that these companies have access to so many startups will probably decrease the amount of competitors in the future," Lopez-Lira says. "Nobody wants to repeat the mistake that Yahoo made by not acquiring Google."

Anchoring your AI startup to Amazon or Google is also a move born of necessity, Lopez-Lira explains. Running large language models (LLMs) at the heart of the technology requires extensive computational power that can drain any startup's budget. Both Amazon and Google have vast resources, and Amazon reportedly controls around one-third of the global cloud computing market. AI startups couldn't afford to run their own tools at scale without the help of such bigger companies. 

"Startups quickly burn through cash and computing power, because at some point, there's [only] so many computers available to run your models in. And that's where the bit players get stuck," Lopez-Lira says. "The conjecture is that every time you use the models, they are making these companies lose a little bit of funding." 

As the tech giants continue to fill a void that might otherwise be occupied by traditional investors, generative AI's principle business problem becomes clear. "It is still not obvious how to make money in this in a generative AI space, it takes a lot of resources," Lopez-Lira says. 

Which is why angling for an acquisition makes perfect sense for a startup with good products, but little in the way of a business model. 

OpenAI, for instance, has not been able to devise a consumer-facing business model beyond subscriptions for its ChatGPT chatbot. 

Ultimately, Big Tech is setting itself up to remain at the top of the pile of the generative AI race for years to come. And many of the startups building LLMs don't particularly mind staying in their shadows, according to Lopez-Lira. The current situation reminds him of Meta's acquisition-spree in the 2010s. Only now, instead of app-based companies being cannibalized by a giant, it's generative AI startups clogging the pipeline. 

"A lot of startups, their pitch is being acquired by one of these giants. Maybe you don't care much about [making a profit] because you're hoping that Google or Meta will acquire you." 

Photo Credit: Getty Images.