There is increasing speculation that AI start-ups are being engulfed by a bubble. This week Mistral AI, a French start-up founded just a year earlier by former employees of Meta, Google DeepMind and other companies, raised $640 from Nvidia and Salesforce, valuing it at $6 billion. That’s triple what its value was in December.
Cognition Labs was launched in November last year and is said to be in discussions to raise money from investors, which would value it up to $2 billion. This is almost six times the value of the company after raising $21 millions earlier this year. Cognition is creating a fully autonomous AI software developer called Devin.
Toppy multiples, coupled with the lack of an end to the exit drought, could cause some VCs to put off new AI bets, if not out of fear that they will miss out on something, which has never really gone away. Touring Capital is an AI-focused venture company I spoke to recently. They hope to protect themselves from inflated valuations, by not investing in hardware or infrastructure at the foundational layers in the AI tech stack. Instead, they invest in B2B applications built on top.
Touring, which is based in San Francisco currently raises its first fund, the Oakley Touring Venture Fund. Priya Saiprasad, co-founder, general partner and Venture Capital Journal , told Venture Capital Journal that the fund is aiming for $300 million with a closing expected later this year. Oakley Capital, 30 founders backed by Touring’s partners and other LPs are included.
The fund’s sweetspot is Series B rounds, which are for AI-enabled software companies that earn $3 million to $10 millions in annual recurring revenues and have figured out their basic product-market fit.
The fund’s size is due to Touring’s desire for a large amount of money to be invested, as well as his desire to “de-risk” and “diversify” the portfolio by making 12-15 investments. The fund will invest in a majority of US companies, but also in those in Europe, India, and Australia.
Saiprasad, Nagraj Kashyap, and Samir Kumar, her co-founders at M12 and fellow GPs, met each other. All three had previously worked for SoftBank’s Vision Fund. Together, they have backed sixteen unicorns and helped 26 companies reach successful exits including Zoom, Kahoot, and Livongo.
Touring has announced that it has made four of its six investments. In September last year, Touring led an $85m Series C1 round of funding for Pixis. Pixis has developed codeless AI technologies that allow marketers to plug-and play AI products without having to write code. The new funding will allow Pixis to develop its capabilities and refine and launch an AI-powered creative studio. It will also enable it to build strategic product and partnership with various social media brands.
Daloopa is another portfolio company. Its AI-powered extraction engine, and modeling copilot, helps hedge fund analysts as well as equity analysts at large investment firms update their investment models more accurately and faster. Last month, Touring led Daloopa’s $18 million Series-B round. The company is moving away from a bottom up approach where it sells to specific analysts to a model that allows Daloopa to target an entire team. Morgan Stanley invested in this latest round and Daloopa has been thinking about partnering up with other large banks to be distribution partners.
Saiprasad stated that Touring is less concerned about backing companies whose software has been “purpose-built” for a specific business user, and which delivers an ROI, such as increasing ad spending or decreasing cross-selling, and a direct, tangible ROI up front. She said that every company must have a data moat or data-based advantage that they can “leverage in a very focused, thoughtful way” and that can help them build their own customer foundation models.
Saiprasad, along with her co-founders, chose a conservative investment thesis because they were wary of the hype that has surrounded AI since the launch ChatGPT at the end of 2022.
Saiprasad explained that “by the time we reach [an investment opportunity in the Series B stage], the software multiples are still used, and not AI. “But our hypothesis is that eventually, over time, you will prove out the AI moat and then hopefully, when you exit, you’ll be at AI multiples, if the market remains the same. If the market crashes, you can fall back on software multiples. We’re not betting on AI markets to remain as hot as they currently are because we’re entering the market at software multiples.