Venture capitalists have eagerly pursued AI deals, with a considerable focus on foundational models.

Despite this trend, it is becoming increasingly apparent that the most promising investment opportunities may lie beyond this initial focus area.

In this article, we feature comments from Daniela Braga, Founder and CEO of Defined.AI, Alex Gurevich, Managing Director of Javelin Venture Partners, and Eric Hsia, Managing Partner at Translink Capital, to delve into their insights and expertise on the topic. 

Is the AI hype warranted?

“This isn't something that happened overnight. It's been decades and decades of building. AI has been going around for 40-50 plus years, going back to Alan Turing, and it's getting better and better and better,” commented Alex Gurevich, Managing Director of Javelin Venture Partners. “I still remember previous hype cycles, Machine Learning, Big Data. Now you’ve got Gen AI, which is a massive leap forward, and the real difference from the prior waves is that you had incredibly powerful models that were in the hands of a few, like computer scientists and data scientists, but with Gen AI, it’s democratised the access of these powerful models into the hands of everyone. And companies are really starting to take notice and starting to use this technology just to be more efficient to further what they're doing.”

In the discussion, statistics regarding the revenue of some of the largest Gen AI companies in the market were brought up, such as Open AI and Anthropic. “I think that's a pretty strong testament that this isn't just hype. And the speed of this adoption, that's mind blowing to me,” concluded Gurevich.

“I think any innovation, any technology, goes in phases. We had the Internet ages in the 2000s, we had Cloud Computing in the 2010s, and this is the decade of AI.” remarked Daniela Braga, Founder and CEO of Defined.ai.

“Those who are building something disruptive and standing out from the noise will succeed and be pervasive. And the thing about AI is that it's pervasive across every industry. This is what makes a hype become a trend, and then become a pillar in the innovation of humanity.”

Which areas are you looking at in AI investments?

Gurevich discussed what he has been seeing within AI investment: “For us there’s been a lot of investment that’s gone into the infrastructure of AI. It’s where a lot of the revenue data has been generated, over 90% of revenue associated with it has been on the infrastructure side: the chip side, the compute side. There’s been a lot of investment going into the foundational model side over the past 24 months.

“There are three areas that we as a firm are spending time looking at. One is AI agents. So not just what can you do with text or images, but from an agent perspective? What can AI do for you, and from a decision process, make decisions, and take its ability to then drive decisions and outcomes. Two is multimodal. A lot of what we've seen so far has been text and image-driven. But if you think about how we all process content, we look at text content, video content, and audio content, so multimodal models will be really interesting, and how can you draw more of the world's information into these models, and build value from that. And the third is in the infrastructure of AI.”

“We’re a software investor. We’ve been really drawn to enterprise use cases. We feel like that’s where you can really develop some differentiation and stickiness,” commented Eric Hsia, Managing Partner at Translink Capital.

“What we don’t like is when founders come to us with an AI label because they use the OpenAI API. We want to seek enterprise companies that are putting a thicker wrapper around these core LLMs. That could be anything from the way you deploy it, security, ethical guardrails, things that enterprises really want to adopt.

“To my prior point about enterprises adopting AI, one example: we backed a company called Resemble AI, which is in the synthetic voice space. They do things like text-to-speech and voice cloning, across entertainment, gaming, and the conversational AI sector. There's a lot of efficiencies that enterprises get with that. But as you can imagine, there’s some legal and copyright IP questions associated with it. So, what they've led with is an ethically first approach to security, first approach to deployment, where companies can, use this technology. What they do, they provide IP protection in the workflow. And they also provide a digital watermarking technology and a fingerprinting technology so that you, as a company, can be sure that whatever models are being used to train this particular voice data, aren't abusing anyone's anyone else's IP. Likewise, if you have some content, and you're worried about others abusing it, you can actually go out and see if anyone is abusing your stuff as well. They also provide things like on prem deployment, and that's something that we've seen enterprises really value and adopt. So, we've been spending a lot of time in the enterprise sector, and I think that's where we're going to be spending more time.”

As a Founder and CEO, Braga commented on Hsia’s view on the situation: “I’m super happy that your firm looks at the ethical side of where the data comes from and if the copyrights are taken care of, if it’s consented, is it paid for? I personally don’t see that.

“Overall, investors look at growth at all costs. They ask very few questions to their companies, and not asking the hard questions about it. So growth at what cost? Is the data paid for? Is it consented? Was it scraped? Do you have all the legal rights of the entire pipeline in the chain? How are you going to address that? [As a founder] I have, but I don’t see VCs asking those questions at all.”

What can set founders apart in the space?

Gurevich said: “As venture folks, we’re looking for the size of the addressable market opportunity. So one is the size of the scale of the opportunity, if you do achieve what you're looking to do, you know, is it big enough from a venture perspective? The second is being at the early stage, we're really focused on you, the entrepreneur. It’s a long, long journey, and we're really focused on people, that's the soft side of our business, and making an evaluation of that, so that's really important to us.”

Hsia added: “We’re really looking for proprietary angles. Whether it’s a proprietary workflow, proprietary algorithms that are complemented by these LLMs, proprietary datasets, that’s what we’re looking for to set you apart.

“For us, it doesn’t feel good to invest in technology that is going to automate people's jobs away. We want to invest in technology that's going to do good for people and lift them up. This is a big philosophical and ethical conversation about the future of AI and what that means for retraining, etc. And that, of course, is likely to happen. But there's also a lot of good that can happen from AI. So, we want to hear that story. We want to hear the story of how it’s going to help people do more, how it’s going to help entrepreneurs build awesome companies. One point that I think is super relevant and critical, when we look at businesses, even if they're not building native AI applications, I want to see how these entrepreneurs are leveraging AI internally to help them figure out what they should do. Because what sets startups apart is that they're able to move faster, they're able to do a tonne of experimentation, do data-driven decision making and figure out what the right answer is. And I think Gen AI can help you do that, that much faster.”

The insights shared by the panel illuminate the ever-changing world of AI investment. It's clear that as the industry develops, the real goldmines in AI might not be where we've been looking so far.