Horowitz took on the then very-early-career Raghuram as a product manager at the seminal dotcom-era browser company. In the decades since, Horowitz went on to build VC giant Andreessen Horowitz (a16z), while Raghuram took his own path to VMware, rising to CEO in 2021, and two years later shepherding the cloud infrastructure giant through its $69 billion acquisition by Broadcom.

This week, the two former colleagues were reunited, with Raghuram joining a16z as a managing partner. Raghuram, who will be a general partner on the firm’s AI infrastructure and growth teams, is joining a16z at a time of great change—both in the VC business and, thanks to AI, across the tech industry. 

I sat down with Horowitz and Raghuram to discuss some of these changes, and to learn more about the opportunities and challenges that the VC firm, and its newest partner, are thinking about.

“We’re moving from a deterministic way of computing to a probabilistic way of computing,” said Raghuram, adding that the AI shift means everything about “what the world of computing should look like” needs to be rethought. 

“Reasoning capability is going to become plentiful and very cheap over time,” Raghuram explained. “Once you fundamentally understand that we’re building for the world of reasoning abundance, there’s opportunity in enterprise, consumer, and across various sectors.”

Horowitz described it as “probably the biggest opportunity set that we’ve seen since we were a firm, as we’ve kind of reinvented all of computing.”

Now 16 years old, Andreessen Horowitz has evolved a lot since its founding in 2009. The VC firm, with more than $40 billion in assets under management, has expanded into private wealth management and public market investing, and stunned Silicon Valley during the 2024 presidential election by coming out as a vocal supporter of Trump. 

AI opens up even more opportunities to invest and to build startups, with the VC industry more reliant on AI investments than ever before. Still, Horowitz stressed that a16z’s vision remains firmly rooted in founders.

“We’re not going to become a private equity firm where we buy a bunch of companies, roll them up, and use AI to, say, fire the employees more efficiently,” said Horowitz. “That’s just not our thing. We’re dream builders, and we want to help people build new things.”

In some sense, Horowitz told Fortune, Raghuram is now filling a gap left vacant by Scott Kupor, who joined the Trump administration to lead the U.S. Office of Personnel Management in April. I asked Horowitz what role politics plays in how he’s thinking about the investing landscape. He emphasized three areas: increased clarity around crypto; his optimism about Trump’s AI executive order; and his hopes around rare earth mineral mining and manufacturing, especially for defense. He also addressed immigration, as controversy mounts around how the Trump administration is handling H-1B visas. 

“The immigration thing is just very complicated,” he said. “Half the H-1B visas have gone to these big consulting firms, looking for cheaper labor, particularly the big Indian consulting firms. That needs to be cleaned up.”

Horowitz authored The Hard Thing About Hard Things more than a decade ago, and I asked him and Raghuram about an observation from the book I’ve thought about many times: that what makes hard things hard is that there are “no easy answers or playbooks.” What are each of them currently thinking about, where there are no easy answers?

For Horowitz: “I would say the biggest one is the relationship with China. There are so many tremendously talented Chinese-Americans, and there are many Chinese nationals in the U.S. who are helping our companies. We do have this tough diplomatic situation with them, both on trade and national security. Right now, we’re definitely not investing in China for obvious reasons. Long-term, how that situation is going to play out and how we should think about it is tricky.”

For Raghuram, he’s grappling with how labor will be affected by AI: “How do you deploy your resources?” he said. “As you lay out people to do a job or AI, what is the right point of the spectrum at that point in time, and how do you navigate it? And if you aggregate that across our companies, and society, how do you deal with that change?”