Why we bet on trust when Anthropic was not the consensus pick, and what the last three years have taught us
Today, Anthropic began trading on the Nasdaq under the ticker ANT. On behalf of everyone at Qualcomm Ventures, congratulations to Dario, Daniela and the entire Anthropic team. It is a milestone few companies ever reach, and fewer still reach this quickly. Less than three years ago, over the Christmas holidays of 2023, we made our decision to invest alongside Menlo Ventures.
On a day like this, it is tempting to tell the story as if the outcome were obvious. It was not. Here is how it looked from where we sat, and why the qualities that convinced us then are the ones that will matter most in the years ahead.
Not the consensus pick
In late 2023, Anthropic was not the front runner in the frontier AI model race. Claude had launched commercially only that March, entering a race in which others already had a head start. Watching the AI industry events unfold throughout the year, however, I came to a conviction that shaped our decision: the AI industry needed a trusted platform, one that could push innovation forward in a steadfast way, with the stability and judgment that enterprises and the public could rely on.
Anthropic stood out as that platform for three reasons. The first was the team. Its founders had been at the forefront of building large language models, so they understood the frontier as well as anyone. The second was staying power. In the months before our investment, Amazon had agreed to fund up to $4 billion and Google up to $2 billion in notes [1], [2], giving Anthropic the AI training infrastructure backing to keep pushing along the AI scaling laws. The third was demand. Enterprises were so enthusiastic about its models, especially for AI-generated code, that management kept raising an already ambitious plan.
Phone calls from the ski slopes
Conviction only matters if you act on it. Qualcomm Ventures was among the first to commit to Anthropic’s Series D led by Menlo Ventures. We ran an expedited investment decision and approval process through the Christmas holiday, with more than a few calls taken from the ski slopes, and the round closed in January 2024. I remain grateful to the colleagues across our team who worked through part of their holidays to make it happen. I especially want to thank Quinn Li, who leads Qualcomm Ventures, for the rigorous debate that sharpened our conviction and for his support in getting this investment done. And thank you to our friends at Menlo Ventures for the great partnership, from that holiday sprint to today.
Inductive reasoning, and a thoughtful answer
In October 2024, I hosted a fireside chat with Matt Bell, Anthropic’s Head of Product Research, at our annual CEO Summit. I asked him about what I saw as a fundamental limitation of large language models: they are, at heart, engines of inductive reasoning, and that can lead to hallucination.
Matt’s answer has stayed with me. Human brains, he pointed out, have used inductive reasoning to emulate deductive reasoning for thousands of years, and with remarkable success. There was every reason to believe AI could learn to do the same.
Looking back, that is exactly what happened. Anthropic pioneered Claude Code and other increasingly sophisticated agent harnesses that let models plan, check their own work and correct course, and in doing so it helped usher in the era of agentic AI.
A name that says it all
Among the many startups I have worked with, Anthropic is one of a kind. It is deeply driven by a commitment to do the right thing for humanity, and that commitment has been hidden in plain sight all along, in its name. “Anthropic” comes from the Greek “anthropos”, meaning “human being.” The word describes anything related to human beings and human existence.
True to its name, Anthropic is organized as a public benefit corporation, with a Long-Term Benefit Trust that elects a majority of its board. True to its name, it pioneered Constitutional AI, giving its models an explicit set of principles and making the alignment of AI behavior with human values a core engineering discipline.
The right path is rarely the convenient one
Doing the right thing is often not the most convenient path. Yet every one of these choices made Anthropic stronger. The mission kept the company laser focused. Trust and reliability became the core differentiator of its models. That focus led it early to the enterprise market and to AI coding, where reliability matters most. And it allowed Anthropic to lead the industry, again and again, to new frontiers such as agentic AI.
Looking ahead
Today is a celebration, and Anthropic has earned every bit of it. It is also a new chapter. The promise of AI is extraordinary, and so are its risks. Navigating both will take exactly what has defined Anthropic from the start: an exceptional team, an ambitious yet deeply personal mission that Dario and his colleagues pursue every day, and the trust they have built with customers, partners, employees, policymakers and investors.
Those are the reasons we invested in the first place. They are also why we believe Anthropic is one of the best guardians to manage the risks of AI while striving for its promise.
To Dario, Daniela, the co-founders and every member of the Anthropic team: congratulations and thank you for letting us be part of the journey. The most important chapters are still ahead.
Source 2:
https://finance.yahoo.com/news/google-bets-2-billion-ai-002731106.html
