# Why investors continue to pour billions into AI

*Matt Murphy of Menlo Ventures is one of Anthropic’s earliest investors. In 2023, his fund invested in the company, which had neither a product on the market nor revenue. Today, Anthropic is one of the leaders in the AI market. At SF Tech Week, he [shared](<https://www.businessinsider.com/one-of-anthropics-earliest-backers-shares-where-hes-investing-now-2026-10>) how the fund decided to make this deal, where it looks for new successes, and what advice he gives to founders.  

***Matt Murphy, Partner at Menlo Ventures**

**Why we invested in Anthropic**

In 2023, Anthropic was valued at $4.5 billion. The company had no product or revenue yet, but we believed in the potential of foundational AI models.

We did not lead the investment round because the company’s valuation did not fit our funds’ strategy. However, we decided to invest a small amount to secure a stake in Anthropic.

We were attracted by the company’s team and technology. Dario Amodei is a strong researcher and technologist, and the team shared a common mission. On many benchmarks, Anthropic’s model outperformed OpenAI, while its development cost was roughly one hundred times cheaper.

Later, we led the Series D round and participated in all subsequent funding rounds.

**How we helped Anthropic grow**

We helped the company hire employees, find clients, and develop a growth strategy. Anthropic also formed partnerships with Amazon and Google, which invested billions of dollars into it.

From the very beginning, the company placed great emphasis on AI safety. This is especially important for corporate clients, who need to understand how reliable the models are and what risks are associated with their use.

AI is evolving faster than many expected. Therefore, it is important not only to improve models but also to test their safety. For example, independent experts can evaluate the capabilities of technologies and the associated risks.

**Where we invest in AI**

With the emergence of open-weight models, the market began to change. If previously companies mainly looked for ways to apply AI, now more attention is being paid to the cost of using it.

Currently, about 60–70% of our venture investments go to AI infrastructure. We invest in computing power, tools for running models, and solutions that help utilize graphics processing units more efficiently.

Among the companies in this area are Modal, Fireworks, OpenRouter, and Gimlet.

At the same time, I expect that over time there will be more applied AI products. Currently, there are more infrastructure startups than companies creating solutions for specific tasks.

I am also interested in startups that apply AI in specific industries. For example, Chai Discovery is working on technologies for drug discovery, Axiom on mathematical problems, and Skild AI on robotics.

**Advice for startup founders**

Do not compare your company with the fastest-growing startups in the market. For a mature business, growth of 30–50% can be a good result if the company has strong financial performance.

It is important to monitor expenses and strive for positive cash flow. Even in a competitive AI market, new companies can find their niche.

In meetings with investors, I advise talking not only about growth but also about gross margin, and how to improve it. You need to have a good understanding of competitors and clearly explain how your product differs from other solutions.

**Why we invested in Factory**

Another investment of ours is Factory, which develops AI tools for creating software.

The company offers not just a tool for writing code, but a solution for various stages of development. Its AI agents, called droids, perform different tasks in the process of creating software.

I also noticed the transition of a specialist from NEA to Factory. The move of experienced employees to a startup can be a signal that the team has potential.
