What is Driving VC Interest in Generative AI?

2022 was not the best year for tech startups, especially those in the software development segment. The world we woke up to after the pandemic was a different one. It no longer had the appetite for software that it had during the pandemic. The Russian invasion of Ukraine further complicated things, driving up prices and triggering an inflationary wave. The Fed was forced to hike interest rates multiple times, the ghost of a recession was looming all over us, and investors took note of all these developments.

What investors think matters because where they put their money serves as an indicator of the near future in tech. The ideas and projects investors pour money into turn into magnets, attracting attention from founders looking to hit big. The concepts no longer backed by investors lose their glamor and start to be questioned by the wider public.

A sea change in investment circles

Once again, investors voted with their wallets and hit the brakes on Web3 projects. Cryptocurrencies and NFT-based ideas no longer engender the excitement they once did among investors. The low-code/no-code segment also saw funds drying up in the last quarter of the year. The segment was running on all cylinders during the pandemic, capitalizing on the rush to digital transformation. With everything going back to normal, expectations were revised in a way that would allow for a more modest level of growth for low-code/no-code companies.

There was one area that seemed immune to the lack of funding, though: Artificial intelligence (AI) and machine learning (ML).

Venture capital (VC) interest in AI is nothing new. Total global investments in AI were less than $3 billion in 2012 and reached $75 billion in 2020. In the U.S. and China, mobility and autonomous vehicles were the top destinations for VC money during this period.

Y Combinator's (YC) first 2023 batch demonstrates that the VC interest in AI is stronger than ever. 51 of the 183 startups in the batch are AI startups, 32 of them focusing on generative AI, that is, the algorithms that can be used to create new content like audio, code, images, text, and video.

This comes on top of the loss of faith YC experienced after its initial zeal in backing AI companies from 2013 to 2017. The first wave of AI tools failed to live up to the expectations, which were probably unrealistic to begin with. Falling short of expectations caused disillusionment and proved that a little bit of restraint was due. Things are looking up once again for AI, and VC funds are pouring in money to power projects of all kinds.

'Promise me it will be different this time'

You probably remember seeing similar levels of unbridled enthusiasm for Web3 and metaverse just a couple of years ago. Are we going through the same things all over again? Is AI another fad that will not amount to much, just like Web3 and metaverse before it? Probably not.

Despite all the excitement it created in tech circles, Web3 failed to deliver practical use cases that can encourage mass adoption and justify more investment. The crypto winter, the FTX collapse, and the never-ending scams killed the short-term prospect that cryptocurrencies would replace fiat money any time soon. The fact that use cases for blockchain remained rather limited and ventured little beyond financial speculation eventually dampened the enthusiasm of investors. The Web3 episode is definitely not over yet, but put on hold until meaningful real-life applications appear.

AI is a more mature technology compared to Web3. Scholars and engineers have been writing and thinking about, and experimenting with different AI applications since the mid-twentieth century. Therefore, it is no surprise that AI proved more resilient because this is not AI's first iteration.

AI has traveled all the way along Gartner's hype cycle: It gained popularity on the back of a major innovation, produced sky-high expectations early on but failed at fulfilling them, falling into the trough of disillusionment. As people realized what AI was good at and how it could be leveraged, this new technology gained a new form in generative AI and pulled itself out of the hole, gathering steam once again with more sensible use cases.

A new wave of AI-powered products

So, AI differs from Web3 in that entrepreneurs have managed to turn it into products that can be used by the masses. Different startups from various verticals like gaming, copywriting, and graphic design are leveraging a model layer like the one Open AI has and building application layers on top of it. Taking advantage of APIs and open-source principles, generative AI startups can quickly monetize their ideas, and investors do not hesitate to back them. Jasper, the AI-assisted copywriting app, provides a case in point. The company grabbed headlines in October with a $125 million Series A round, becoming the poster child of this new generation of AI-powered apps.

The VC interest in generative AI was almost palpable in the last few months of 2022. There were 78 successful funding rounds closed, with a total investment of close to $1.4 billion. Besides Jasper, Stable Diffusion and Runway topped the list of companies raising money at incredible multiples despite having modest revenue streams as of now.

However, the niche that gets VCs drooling consists of startups working to develop new foundation models, the large AI models trained on a vast amount of data for use in a range of implementations, which can rival the ones like OpenAI's GPT. One such company is Cohere. The Toronto-based company is looking to raise money at a valuation of more than $6 billion. It stands apart from OpenAI with its focus on business-applicable use cases and its target customer base of enterprise users.

Companies like Cohere are the innovators that can broaden the horizons of AI through research and development rather than building "me-too" apps that can attract a few investors before flaming out. The hope is that these efforts will produce new models that have less bias and consume less energy.

Conclusion

Angel investor Elad Gil was quoting Andy Rachleff when he said, "the market impact, good or bad, dominates the team impact." The VC interest in AI is so strong and the initial reaction from the public is so positive that generative AI startups enjoy a tailwind these days.

However, this will not last forever. The laws of startup survival will still hold at the end of the day. Once the initial VC interest subsides and profitability becomes a priority, the startups with the better team, product quality, traction, and moat (defensibility) will prevail. In the words of Warren Buffett:

"Only when the tide goes out do you discover who's been swimming naked."

Startup founders in the generative AI segment had better start looking for some semblance of swimming trunks if they don't want to embarrass themselves or, worse, lose their startups.