In a Masters of Scale Summit interview, venture capitalist Vinod Khosla outlines his "Instigator" thesis: solving the climate crisis requires a dozen breakthrough entrepreneurs—one in each of the twelve largest carbon‑emitting sectors (electric vehicles, aviation fuel, plant proteins, steel, dispatchable power, fusion, etc.). He cites Elon Musk (EVs), Pat Brown (plant proteins), and Bob Mumgaard (fusion) as exemplars of instigators who shift paradigms and catalyze whole ecosystems. Khosla predicts that by 2030 eight of the twelve sectors will have viable, scientifically sound solutions, with large‑scale deployment in the 2030s. He is skeptical of current AI climate work, arguing most efforts target minor efficiency gains, but foresees a future of "AI scientists" who dramatically accelerate material design and fusion plasma control (e.g., Convoy Fusion’s partnership with DeepMind). Extending the thesis to healthcare, he argues startups, not big pharma, will drive radical change, especially through AI‑designed, personalized drugs, though regulatory timelines remain a barrier. Khosla also shares his broader vision that AI will make most expertise near‑free—AI doctors, therapists, engineers, marketers—transforming labor markets. He acknowledges massive job displacement (up to 80% of jobs) but predicts a shift to abundance, with GDP growth rising from 2% to 5% and a future where work is chosen rather than forced. Politically, he labels former President Trump as a negative instigator, but expresses optimism that rational, inclusive values will prevail. He refines his techno‑optimism with "care and caring," emphasizing AI safety and societal inclusiveness. The conversation blends personal anecdotes, strategic forecasts, and philosophical reflections on humanity’s future.
Chapters
Good afternoon or evening, everybody.
Good afternoon, good evening, and good night. Um, uh, will you start by telling us about your Instigator thesis?
Um, I first came up with the idea of Instigator. I've dealt with instigators, which are really entrepreneurs. All my life. But when I was looking at the climate problem and people kept telling me it was a very hard problem, not a solvable problem, I looked at it carefully and I said, there's really, it's a much simpler problem than people make it out to be. You only need a dozen instigators. In a dozen areas. I don't buy the thesis, you know, everybody can help, everybody can contribute, but we need a dozen instigators to solve the climate crisis.
Well, I mean, okay, so it matters, but it doesn't matter in the grand scheme if I drive an electric car or what have you. Why a dozen in a dozen areas?
Because there's only a dozen large areas that have large carbon emissions. Now I'm talking carbon emissions. I'm separating climate change from sustainability. To me, they're quite different things, though they're often confused.
Can you unpack how you view them differently?
Sustainability is about recycling and things like that. Green approaches. It gets mixed in with some of the wellness approaches to food and other things.. But if you are strictly looking at the climate change problem, it's about carbon emissions. And there's a dozen important areas. Electric vehicles is one. Aviation fuel is another. Plant protein or proteins is another. There's only a dozen large emission areas that really matter. If we solve them, We will mostly have solved the climate crisis or the carbon emissions crisis. If we don't, it doesn't matter what else we do, you know.
And your view is literally one instigator, one entrepreneur in each of these categories can make a difference, can make a difference or can make the difference.
You know, in— take what I hate giving Elon Musk any credit, but we've been battling. But I do give him credit for instigating the change in to electric vehicles. Without him, we would be on a very different path, dependent on General Motors and Volkswagen, Ford to ship electric cars. And frankly, I looked at this area pretty carefully around 2010, 2010, the Department of Energy made a forecast for the number of electric cars in 25 years, the year 2035. And Elon Musk exceeded that forecast in 2016 or so. So he did more than without ever being in the car business to change the paradigm of what's accepted and then got everybody to follow.
We contain multitudes, even Elon Musk. Yeah.
Yes. Um, you know, Pat Brown did that with plant proteins, uh, and I think those are well on our way. They'll have up and down cycles, but I think we are well on our way to solving the protein problem, and it's through fundamental science innovation. Um, Bob Mumgaard's done that in fusion. I don't think fusion's, uh, If question, I think 5 years from now, nobody will be debating whether fusion is possible. It'll be how fast can we implement it.
Why do you believe we'll be there within 5 years?
Just because, you know, if I looked at 2018 when we— I think that was the year we invested in Convoy Fusion, by the way, also invested in OpenAI the same year. And we can come back to that. And a public transit system. 3 fundamental things in 1 year. I was very proud of that year. But it isn't that Bob started that company. We started it in a way where in the following 5 years, maybe a dozen, maybe 20 startups started doing the same thing. And then when there's that many shots on goal and talent and money are flowing in, which is what Bob caused to happen, and that's an instigator. So at this point, whether Bob succeeds or fails, I think the world will have fusion because there's so many good scientific attempts.
And do you believe we have enough time to get to all 12 categories?
I didn't speak to that issue. Um, I do think we are well on our way in steel. I think we are well on our way to sustainable aviation fuel. And these are things most of you probably haven't heard of, but I've seen the state of development to say at this point it's mostly engineering risk. Very often the risk is cost, but dispatchable power generation is a big one that there's more than one approach that will work there. And we've instigated that change. Uh, if I take these dozen areas, I'd say 8 of the dozen have very good, scientifically sound efforts to develop the technology. That's the optimist in me. My bet is by 2030, most of these areas will have a solution. Um, and then the 2030s will deploy them, build 5 plants, 10 plants. But at that point, we get to a point where we can do thousands of plants a year in each of these areas, whether it's steel or power generation or others.
I think we are now legally required to talk about AI in every conversation. So I'm going to not break the law here. How will AI help us achieve the 12 on 12? Yeah.
So the first thing I would say is we should be cautious about ascribing too much to AI. How many people know the NeuroIPS conference here? Raise of hands. It's sort of the go-to conference for AI. Very few hands up. And I looked at last year's papers and the years before on AI and climate change. And I said, none of these approaches, these efforts of AI in climate change are going to be consequential.
Why not?
Just because they address minor problems. You know, building efficiency isn't a climate change problem. You can improve it 2-3%, but you're making very little progress unless you get to 80-90% reduction in carbon in every one of the major areas. We are not going to get there. And so I'm only looking for the 90% reductions, not the 10 or 20% improvements in efficiency or other things.
Which categories do you— in which categories do you think AI may help us the most?
Well, so I think in the next couple of years we'll get to an area where we have what I call, and you know, this is not an original term, AI scientists. Then you can apply a much faster rate of scientific progress In materials design, for example, we are using AI in Convoy Fusion to control plasma. Plasma control is a hard problem. We are using AI. Convoy Fusion is partnering with DeepMind to do that. But I do think fundamentally AI scientists to get creative about new approaches in some of these areas will make a huge difference.
I want to take us beyond climate, if I may, unless there's something you want to— No. You want to add there. Taking the instigator thesis into healthcare, for example, we have at least a couple of companies here, including Transcripta, that are using AI to discover new applications for existing drugs, to discover new drugs. How do you think about the instigator thesis as it applies to healthcare, particularly with AI health?
And I think I would broadly say entrepreneurs cause most of the radical change. When it comes to incremental innovation, big companies are relatively good at it. And Intel can do a good job of moving from a 7-nanometer semiconductor process to a 5-nanometer process. Uh, but in the 40 years I've been doing innovation now, only done technology-based innovation, I can't think of a single example of large innovation that was done by a large player or a large institution or a government program. But in each case, whether— I like to put it the following way: did Amazon reinvent retailing or was it Walmart or Target? Did— we talked about Elon Musk. It wasn't General Motors or or Volkswagen. In Airbnb, and it's unfortunate Brian can't be here, you might have thought that, hey, Hilton or Hertz or Hyatt would know how to do that. Or in Uber's case, you'd think somebody like Hertz will reinvent that area. There isn't a single example I can think of a large innovation that a large institution was part of doing or predictable.
So you believe that in healthcare it will be the startups that—
Well, vaccines was done by startups. Now big companies help, they pile on. Today we are dependent on the big company, auto companies to build EVs. But they've all switched because somebody has shown the way. And that's what instigators do. And that is happening in healthcare. Vaccines were done that way. Almost all the large innovation is happening at many levels. You talked about drug discovery. You know, the idea of first drugs being designed by AI. Very powerful. Not a single company, pharma company, adapted it 10 years ago. I think I first talked about it 10 years ago in a long piece I wrote. But every company is now getting on that bandwagon. They'll have—
And how quickly do you think we'll see the revolutionary change, not the evolutionary change, but the revolutionary change in healthcare?
Healthcare depends on regulatory control, so that limits what we can do. At the drugs level, no matter whether you have AI-designed drugs or other traditional drugs, you've got a 10-year process of, of getting a drug approved through the FDA. Now there's a cheat to that that I'd love to see happen, and we are working on it. If you design a drug that's for one person— you know, today we design a drug and it works for all 7 billion people on the planet, whether it's a fancy drug or a simple aspirin. And different people have different genetics. We don't optimize for that. You could design a drug for only one person. I don't know how the FDA would handle that. We are attempting to see how that might go. Design the drugs with the help of AI for one person, one patient at a time. Much easier to do in certain classes like genetic diseases. You can do that. The other end is— and I'm very excited about this idea. It's hugely disruptive. How many people have heard of a thesis I put up on AI about 2 weeks ago called AI Dystopia or Utopia?
Everyone, raise your hands, please. No, you don't have to.
Um, I postulated that all expertise will be free. So if you look up in TechCrunch, I know we are not at TechCrunch, uh, in 2012, I wrote a blog called "Do We Need Doctors?" under a simple thesis in 2012, later expanded to a 100-page thesis by 2016, that we won't need doctors for their expertise. Because AI will be the best primary care doctor. We are working on that. AI will be the best and easily scalable therapist for mental health. AI will be the best oncologist in most of these areas. By the way, it's not restricted to medicine. AI will be the best structural engineer. AI will be the best salesperson or marketing person. So almost all expertise will be near-free. Okay.
The best marketing person, the best salesperson.
Yes.
I mean, that's not necessarily intuitive, at least not for me, because there's—
And there's only a dozen startups doing salespeople, AI salespeople.
So, but in each case, there are sparks of creativity, there are relational components, right? If you and I are working on a deal, there's trust that we build over time and I'm able to ask you questions, you're able to ask me questions, we're able to work together in the best case to create a deal. Are you imagining two AI agents putting that deal together? How are you imagining that being a more efficient and effective process?
The first thing I would say is, Just because I can't precisely predict how the change will happen doesn't mean the change will not happen. I'm almost certain the change will happen and radical change, but it's hard to predict exactly how. So if you were to bet and know in 15 years things will be the same, I'd take 10 to odds on that with anybody.
On every category?
In most categories.
What's an example of a category where you wouldn't?
Where I wouldn't? Look, there's certain things that are not rational that make us human. And I think those categories will be categories. Look, even today you can get a well produce glass like this off a factory line, but I'll pay more for a handmade glass. We all do. I do. And so that's the humanness factor. We prefer certain things and we don't have to be rational about everything in our lives. We like to think of ourselves as rationalists. I do. But there are things I prefer.
And perhaps the irrational decisions are some of the most important ones that we make.
In fact, there's some of the most important decisions. I think there was a scientist, and I won't name the name only because I haven't been able to verify it, who said, "We make most decisions rationally except the most important ones." Who do we marry? That's not a rational decision. That is an emotional decision. And it should be. That's where humanity comes in and humanness comes in.
Yeah.
And I postulate in my thesis that we will have more room for humanness because we won't be occupied by the things AIs can do for us.
I'm looking forward to your next set of articles on the other trillion-dollar categories where AI has the potential to change the world. We've spent a non-trivial amount of time today talking about building a more perfect union. And I can't let you leave us without asking about, well, one, frankly, thanking you for being very public in your political views this year, which not every business leader has.
Thank you. And probably not the smartest thing to do because from MAGA experience, extremists. I get a lot of hate mail and threats.
But thank you for your irrational support of the right thing. Um, uh, uh, my family isn't happy about it. Highly rational and irrational at the same time. Um, uh, how, how does the instigator thesis apply to American democracy?
Well, so you can instigate good things or bad things. And frankly, probably in America, if I look at this, the most important instigator in the last 100 years in the social domain has been Trump. Now, instigator of bad things. He's given people, in my view, permission to go back to their base instincts. No migrants, no foreigners. No affirmative action. No, it's okay to think that way. And I think we are seeing this emergent 25, 30% of the US population that had base instincts that I would consider bad by modern society standards, that he gave them permission to think that way and be, in fact, offensive about those wanting those things.
What's the cause for optimism?
You know, in the political domain, I think we'll have— I have to believe rationality will prevail. I'm an independent. I'm not a Democrat. I'm not a Republican. I used to be a Republican when fiscal was important. After climate became important, I became an independent, mostly to focus on climate as my biggest issue. One of my fights with Elon, I said to him, here's my set of values on who I vote for. Number 1 is values. Obviously you don't think that's important. My number 2 is climate. My number 3 is economics. And then there's 10 others. But a clear prioritization of what my political priorities are appropriately weighted, I think enough people will sign up for the right middle-of-the-road path to rational, inclusive behavior. Now, one— and we should end on this note. I know we are sort of out of time. You know, I'm a techno-optimist. Mark has talked about it, but I modified techno-optimism with a simple phrase which I use in my long— it's a 25-page essay, so not easy reading. I say techno-optimism with care and caring. And the care part is AI safety. And the caring part is inclusiveness about everybody else. And I do think techno-optimism by itself is not necessarily good for society as a whole. It's good for the technologists. But with care and caring, I think it becomes the single most powerful tool we've ever had to improve, to move society to an era of abundance and being more human. Now, people ask me a simple question like, Won't it displace jobs? Honest answer, yes. In fact, I think 80% of all jobs. So if you do the math, 65% of all jobs in 20 years can be done by an AI in those countries that choose to do so. And it'll be a country-by-country political choice. So a lot of job displacement and probably a pretty disruptive period, which I define in my essay. But today, being a farm worker, having to work in Salinas in 100-degree heat 8 hours a day for 20, 30 years, or working on the General Motors assembly line mounting a tire on a car for 30 years, 8 hours a day, That's not jobs. That's servitude because you need to make a living. I'm pretty certain we will have enough abundance, and I postulate GDP growth will go from 2% to 5%. And that kind of growth rate, uh, maybe not our kids, but our kids' kids will live in an era of abundance. And we will have enough resources to support everybody at the minimum standard of living. So people work on the things they want to work on, not the things they have to work on to support their family.
