Talk Gen AI: Investment Trends in Generative AI

Generative AI is the hottest topic in technology. There have been record amounts of investment in Generative AI — in the model providers, tools and services, as well as platforms and applications leveraging Generative AI.

At Talk Gen AI, I had the opportunity to discuss current trends with a panel of investors:

• Divya Sudhakar, Partner at Geodesic Capital
• Jeremiah Owyang, General Partner at Blitzscaling Ventures
• Kristina Serafim, Founding General Partner at Untethered Ventures
• Ryan Floyd, Co-founder and Managing Director of Storm Ventures

Check out the recap and watch the video below.

Investing from infrastructure to the application layer

The panel represented a mixture of small and medium-sized funds investing at the seed stage to later stage growth, across the technology stack from infrastructure to applications.

Kristian Serafim, of Untethered Ventures is looking more at the infrastructure layer — focusing on healthcare and enterprise enablers, especially in data management and security.

Divya Sudhakar of Geodesic invests at the growth stage — Series B or later — across the board from infrastructure to application. However, most of the focus has been on the infrastructure — looking at developer tools and data.

As part of a seed-stage fund, Jeremiah Owyang of Blitzscaling is focused more on the application layer — AI agents. His team looks for companies that can quickly scale through viral effects, product-led growth, and network effects.

Ryan Floyd, of Storm Ventures, also invests at the application layer. They are looking for startups who are rethinking entire workflow processes, rather than just adding AI to traditional applications. While the latter can be a good strategy for an incumbent, upcoming startups will win on completely rethinking the workflow, he explains.

Data is gold

One thing the panelists all agreed on, was the value of data.

Having access to high quality data, and data rights, is a unique differentiator for a startup, as Serafim states. Floyd adds that data is probably the single most important thing in regards to differentiation when evaluating a company.

The data can be valuable in many ways, including being used to train models to develop applications that larger enterprise customers will adopt.

This concept of “exclusive” data is what Owyang’s team is looking for. As he explains, 80% of the world’s data is behind a firewall, not yet trained on by an LLM. Startups that have access and rights to this data, will have a “moat.”

Understanding the startup’s legal rights to the data is an important part of due diligence. As Sudhakar explains, her team dives deep into the contracts to understand what rights the startup really has. It is especially important given regulations, like GDPR in the European Union. She adds, savvier startups build access rights into their contracts — even offering price advantages for access.

Fundamentals are the same

The Generative AI space is moving so rapidly. The major providers are launching new models on nearly a weekly basis.

Each time a new model is launched, it seems as if a whole wave of startups may be put out of business.

At the same time, Generative AI is leveling the playing field — everyone has access to the same models for the most part — both startups and enterprises.

How does one differentiate?

While there are some concerns around defensibility that Gen AI raises, the panelists, for the most part, are following their traditional approaches to evaluating startups for investment.

Nothing has really changed, as Floyd states. The criteria are the same.

Even though Generative AI is hot right now, it is a natural evolution of what they have already been seeing — from big data to machine learning to AI and now to Gen AI, as Serafim explains.

Investors are applying the same evaluation criteria they always have been, as Serafim illustrates. Is the company solving a big problem in a differentiated way, in a big market? Do they have a defensible moat? Can the team iterate fast, adapt, and pivot?

The team is an important factor. Given everything is moving so rapidly, can the startup hire the right technical talent to understand where the ball is moving, and be able to adapt quickly, to be successful and compete, Sudhakar explains.

Owyang adds that his firm looks to see if the founder is an “infinite learner” — someone who can learn new things really fast, and pivot.

Startups need to have product-market-fit. In the case of Storm, that means having at least $500k ARR. For Floyd, it comes down to identifying startups that have an efficient go-to-market strategy. They are looking for those fundamental business indicators.

Watch the video

Arte Merritt is the founder of Reconify, an analytics and optimization platform for Generative AI. Previously, he led the Global Conversational AI partner initiative at AWS. He was the founder and CEO of the leading analytics platform for Conversational AI, leading the company to 20,000 customers, 90B messages processed, and multiple acquisition offers. He is a frequent author and speaker on Generative AI and Conversational AI. Arte is an MIT alum.