Michael Stewart, EWMBA 15, did not set out to be a venture capital investor.

A scientist at heart, he expected to spend his career on the technical side of things doing hands-on research—a career that led him to Applied Materials, where he spent nearly a decade working in semiconductor process engineering and later in solar and cleantech. 

“All I wanted to do for the rest of my life was work on clean energy,” said Stewart, EWMBA 15, at a recent Dean’s Speaker Series talk, in conversation with Gonzalo Vásquez and Rey Yruegas Almanza, both MBA 26. “I came to Haas to get to the bottom of how the various levels of that market might work.”

Watch the video of Stewart’s talk:

Now a managing partner at M12, Microsoft’s venture fund, Stewart leads investments in generative AI, gaming, and systems. He was recently named among the top 50 Emerging Leaders 2025  in corporate venturing by Global Corporate Venturing. His work includes evaluating new technologies, advising founders, and identifying where emerging innovations align with Microsoft’s long-term strategy.

Stewart said he quickly saw that Microsoft would be deeply involved in AI investment. “But also, it led me to question a number of times, including today, ‘What are we doing as VCs?'” he said. “Are we catalyzing something new, or are we just tacking on to a phenomenon that’s already very well capitalized? What’s the different strategy that we can bring to it?’ So it’s still something I wrestle with.”

Stewart’s technical training—he holds a PhD in chemistry—became an unexpected advantage in venture early on. Habits he developed throughout his academic and engineering career helped him to critically analyze complex technologies, recognize underexplored opportunities, and evaluate whether an innovation could scale.

While completing his MBA program at UC Berkeley Haas, Stewart was still working full time at Applied Materials, which was preparing to merge with Tokyo Electron. The cleantech venture downturn was well underway as the VC landscape pivoted toward software, prompting him to consider a new direction after graduation. 

He told his managers he planned to leave after graduation. But after they asked if he would spend three months with Applied Ventures with no guarantee of a full-time offer, he agreed. When the merger ultimately fell through, a role opened on the venture team, and Stewart decided to stay on full time. Even then, he expected the experience to be short-lived, with plans to spend a year or two in venture before transitioning back to the technical side of his career. 

As he settled into the role, he realized that his skills as a researcher translated to his work in venture.

“You have to do a lot of soul-searching,” he said. “It’s that hunger for: What are people not doing?”

At M12, that same instinct for finding gaps shapes how he evaluates companies. Venture often comes down to judgment—being a “good gambler” while also thinking like an inventor, he said. With that unpredictability comes a constant negotiation of risk and timing: Some technologies appear before the market is ready, others after momentum has moved on.

“What are we doing as VCs? Are we catalyzing something new, or are we just tacking on to a phenomenon that’s already very well capitalized? What’s the different strategy that we can bring to it?” he said. 

Stewart offered advice to students about going into venture capital, tapping his decades of experience. “Any of you who are thinking of a career path and then (pursuing) VC, I think that’s still the best way to go,” he said “Get as much experience as you can in the dimension you can be the most creative in.”

Listen to the Dean’s Speaker Series Podcast episode:

Transcript below:

JENNY CHATMAN: Good afternoon. Hi there. I’m Jenny Chatman. I’m the dean at the Haas School of Business. Welcome to the Dean’s Speaker Series. I am so thrilled to welcome back to campus, Michael Stewart. He’s from our Evening & Weekend MBA Program, class of 2015. He’s a managing partner at M12, Microsoft’s investment fund. Super interesting. He’s led 16 investments in generative AI, gaming, and systems. He sits on several boards of his portfolio companies. And Michael’s been recognized for his work at M12 and was named one of the top 50 emerging leaders in corporate venturing for 2025. Congratulations. Amazing. Yeah, applause. Previously, Michael worked at Applied Ventures, leading 12 investments in AI and machine learning hardware and software, silicon photonics, robotics, and printed electronics. Of course, he has a PhD in chemistry or a degree in chemistry. Michael’s also been on the other side of the table, co-founding a consumer electronics focused startup called JUSE, J-U-S-E, in 2014. He also led technology development in the semiconductor industry and is an inventor authoring more than 40 patents. He does hold a PhD in chemistry from Purdue University and an MBA from our Evening & Weekend Program, as I said. We have many students here, Michael, who are very interested in venture capital. In fact, over the weekend, we had a great networking event on Saturday. So we all can’t wait to hear your insights. 

Let me give you a little direction. If you haven’t been here before, there’s a card on your chair. If you have questions during the interview segment, please write them down. Please put your name and program. Carrie and Sarah will be picking them up, and we’ll get to as many of them as we can at the end of the session. Let me now turn it over to Gonzalo Vásquez and Rey Yruegas, who will lead today’s discussion. Thanks. 

GONZALO VÁSQUEZ: Hey, Michael. I’m Gonzalo. I’m a full-time MBA student, Class of 26. It’s an honor to be here with you. 

REY YRUEGAS ALMANZA: Yeah, Michael, thank you so much. My name is Rey, as mentioned by Dean Chatman, full-time MBA student from 2026, and we’re going to get really into it really quick, Michael. 

MICHAEL STEWART: Excellent. 

REY YRUEGAS: You transitioned from being a scientist and an engineer with 40+ patents to becoming a deep tech venture capitalist. What inspired that shift, and how did your time as a researcher influence your investment philosophy? 

STEWART: Yeah, and so, the cliche—everybody gets into VC in a different way that’s really irregular—definitely holds true for me. When I was here finishing up my MBA, I was still working in the tail end of what was photovoltaics and solar energy. For me, it was a really big passion. There was a moment when I really felt this is all I wanted to do for the rest of my life was work on clean energy. And I came to Haas to get to the bottom of how the various levels of that market might work. The technology I had a pretty good understanding of. I had moved from semiconductor process engineering to creating crystal and silicon photovoltaics. It isn’t really that massively different that there’s a huge difference in the cost, the market, how it’s distributed, project finance, all of that was stuff I learned here. And it hadn’t really—the venture market hadn’t really turned around by that point. This was 2012, 2013 time frame. “The Software is Eating the World” article by Marc Andreessen had only been—I think that was 2012 perhaps. So at that time, I mean, it was still news that B2B, SaaS, and VC had found a new record groove to slot into and to start growing companies. And the cleantech venture experience by that point was already starting to look pretty bad. So I did a little bit more thinking about it. And the company I worked for, Applied Materials, had been undergoing a merger with Tokyo Electron to create this company called Eteris. This is 10 years ago. And the venture fund for Tokyo Electron was going to merge with the Applied Ventures team. I had already been at Applied for 10 years. By that point, I don’t know, racking up some accomplishments of what I did in process engineering. 

And I thought I should do something different after my MBA and leave. And so, I told my managers I’d be looking to leave Applied Materials, and they asked, ‘Would you like to spend three months working in the venture fund to apply what you had learned about finance and structuring companies and stuff? We could use your help in these cleantech deals that we’re still doing.’ And I thought, OK, but then they said, ‘There’s not going to be a job at the end of this. This is just for you to learn for three months.’ And I worked for three months, did a quick deal, and printed electronics. The manager of the Applied Ventures team had some extra headcount for this merger with Tokyo Electron that didn’t happen. And so, it was like, ‘Do you want to just join full time?’ And that was 2015. And I thought I would seriously do this for one or two years because I still wanted to be an inventor and an engineer. And I didn’t find the beauty in the investing, in underwriting piece, until I had done it for a year or two. That said, here I am, and now I can really—it’s very hard for me to imagine any other job. 

VÁSQUEZ: That’s awesome. Well, moving forward. It’s not that we want to flex here, but a couple of weeks ago we spoke to Arvind Srinivas. He’s the CEO of Perplexity, and he told us that moving from PhD to founder and raising capital had some similarities. So have you found your path as an MBA or as a PhD to have some of those commonalities? 

STEWART: And how many of you are PhD students out there? Just a few of you. OK, yeah, I mean, when you’re doing a PhD, at least when I was in school, the bar to finish up is you got to contribute something. Come up with something that’s an actual advance or move the field forward. And you have to do a lot of soul-searching, not just about, ‘How do I become competent in what I’m studying?’ but also, ‘Where is there a gap in the field where I can add something to it?’ And I think if I trace back to that one piece is the No. 1 commonality between being an investor and being a PhD researcher. It’s that hunger for: What are people not doing? Where is this territory that could be uncovered or developed that other very, very smart people have missed? And in some of the times, you’ll see an area that’s just so vast that you’ll doubt yourself, like it can’t possibly be that people are not seeing this. We’re actually still experiencing that to some degree system wide with AI. So as someone who’s driving your own research, you have a PhD adviser. You could have the best PhD boss in the world, but it’s up to you to really look for something new and then figure out how to find a way to be part of it. Yeah, that’s something I’m still trying to do right now in this job. 

YRUEGAS ALMANZA: Good morning to venture capital. A lot of our fellow students are interested in that career path, but there’s also CVC, corporate venture capital, which is a branch of it. So would you mind giving us a quick masterclass on what that is, and how do the companies set up those funds? What are the expectations? What are they trying to achieve that they cannot do with a traditional PC? 

STEWART: Yeah, great question. Why CVC? Why invest? Why have a venture fund? Almost every company that has a CVC arm also has a corp dev team that can do—presumably, they could hold equity. They could do investments like a venture fund. They could do—when done well, the venture fund acts as an outpost that operates in the venture ecosystem exactly as a VC, a financial VC, would. And in today’s reality, there are fewer and fewer clear delineations about what types of funds could be grouped with other funds. Everything from super early pre-seed check writers who are just individuals, and then they create a fund based on their own investing to specialist seed funds, to multistage seed through series C, and then growth. It’s all run together. 

So what we do, I think, at M12 that’s exceptionally good is, one, we’re not trying to tackle every deal out there. We’re not trying to blanket the ecosystem with investments in half a dozen companies all competing in the same space, just because they might do business with Microsoft. I feel that would be a very trivial use of what we have. It’s a little bit of a hybrid between spotting early companies, providing them with the expertise, advice, connections, and cash that they need to become effective in the ecosystem as companies selling things, making revenue. And if we’re early enough and in the right spot, Microsoft can come in and aid them as a partner way earlier than would naturally be possible. Microsoft will work with you. If you guys were founders of a company as large as Perplexity now, yeah, it’d be no problem. I mean, Microsoft wants to do business with you. If you are a million, $2 million, $5 million revenue, it may not be possible to really get the full help unless you have people underscoring and building the strategic rationale for it. And that’s what we specialize in. 

YRUEGAS ALMANZA: One question. So unlike traditional VC funds, M12 is funded annually from Microsoft’s balance sheet, from our understanding, rather than raising the traditional 10-year fund. How does that structure change your decision making, which I think you touched on a little but also what happens during a year when Microsoft faces some headwinds? Does as M12’s capital allocation shift, or is it relatively insulated? 

STEWART: Oh, yeah. So I can answer a little bit about that. We don’t really talk about the fund dynamics or mechanisms in public, but many CVC funds are just balance sheet CVC funds, meaning there’s no fund structure. I mean, you could ex-post facto make one through grouping the transactions as a fund. And then, there are some that raise with the main LP and outside capital. They raise basically an outside fund and operate as an LLC. 

So we’re a little in the middle. We work for Microsoft. We invest their balance sheet money, but we conduct a fundraising process with Microsoft. And we operate annual fund vintages. It’s a little bit better than the evergreen model of just there’s equity in a pool and it could be any size. And the amounts going into the fund go up and down year over year. It’s harder to do that and stay consistent in your internal process if you have one versus the fund life helps create some rigor around the check sizes, the amounts, the follow on process, and also the stage we deploy in. 

YRUEGAS ALMANZA: Michael, super timely. We’re taking a VC class, and we recently spoke about corporate venture capital. Our professors told us in a way that when speaking about CVCs, it’s good to see 100% financial, 100% strategic, and most of them are really in a spectrum. I’m curious to understand how you think about Microsoft in that spectrum, and if you agree with that spectrum, or should CBC be thought of in a different manner? 

STEWART: Should it be totally binary? Yeah, no, it’s always some place in the middle. The reason why is you have some funds that will operate completely ignoring what the parent company wants or the main LP wants. And I’d say to that, just go out and find me a financial venture fund that’s super successful that ignores their LPs. That just doesn’t happen. Limited partners are limited partners. They’re not directly intervening in a deal process. They’re not saying, ‘Whoa, you’re about to do this investment. Hold up.’ That’s statutorily, they’re separate from the fund. And yet the dog and pony shows—and including LPs and meeting companies and sometimes granting freedom to invest. You’re doing all that because they’re giving you money, and you should appreciate it. Any of you that do get into an institutional investing job, don’t forget where that money is coming from. I mean, be serious—like, it’s someone else’s money, not yours. So we keep that in mind. And we also keep in mind the reputation of Microsoft as an extremely important thing that could be squandered if not damaged if we’re careless. So everything we do around strategic alignment is self-policed. It’s the leadership of my fund that enforces that. We enforce that in all check writing operations we do. And I usually start thinking about an investment with, why would this matter? Is this the company I could bring to Satya and the staff at that time, at the time when that would be appropriate and they would find it really interesting? Because they have visibility into nearly everything worldwide. That’s the other force to consider. If we were investing on behalf of a company who had no real name recognition in Silicon Valley, much less worldwide, and the goal is to elevate the identity of the brand, we would have totally different tactics in what kinds of deals we deploy into. 

YRUEGAS ALMANZA: Just a quick follow up. We spoke about it briefly in the green room. But what’s your way to approach knowing where Microsoft is heading, what people at Redmond are thinking so that you think, all right, these are the type of company or technologies that we should be looking at? 

STEWART: Yeah, the first and easiest way is just to ask them. So I just did a few weeks ago my showcase, our showcase with Satya and the executive staff. And again, this is like a two hours worth of time. I bring some startups to meet them, and you can imagine with the CEO of Microsoft, all pretty much of the C operator officers in a room. And I’ve got them for two hours. That’s expensive time. So I can’t really use that to be like, ‘What should we be investing in, and what’s cool?’ So you do a lot of networking inside a company, and you’re doing it with people who, one, have an inclination to see more startups, but who are working on a product. They’re working on a business strategy. Two, they want to see more startups. And this is the other dependent level. They know what they’re looking at when they meet a startup. 

So one of the things I noticed about Microsoft product people was that—yeah, I mean, sometimes they’re meeting a startup and they just look at this like I don’t get what’s special about this because they turn around and they’re running seriously like a $10 billion business. And they’re just—they’re one of a dozen people doing that. So you can’t always just go straight to the top person and look for their stamp of approval, their imprimatur. You’re not looking for them to just say, ‘Here’s a check mark. Now go do the deal and ignore everything else.’ You’re wanting to think like, say, ‘These guys are working with us. Would it be a good fit with this other program? I may know about that; it could be secret; it could be public. But they’re going to need partners. Or they might do business with a startup in a way that is much larger than the ACVs that they have right now.’ That feeling it out process takes a lot of—it just takes being available and being part of the internal community of the company. 

VÁSQUEZ: And just to make sure everyone here today understands the difference between CVC and VC. What can you do as M12 that Sequoia or a16z cannot do? And conversely, what can they do that you wish you could if anything? 

STEWART: Well, I mean, if I’m speaking, like what I just said, there’s a business area or I’m using the word gap. You guys what I mean with gap? If I said Microsoft wants to sell more Copilots. And I’m just looking for sales tactics that help them meet a quarterly. That’s not a very interesting place for a venture capital fund to be. We shouldn’t be addressing this quarter, this year, even this two-years horizon. It’s instead thinking about—I’ll give you a theme I’ve been working on the last year and a half is really the data center of the future. Microsoft is already the leading operator of this class of data centers worldwide. I think it’s like 430 data centers or something. Don’t quote me, but it’s a large number of data centers that they own and operate. And in the time that I’ve been investing at M12, something is happening to this industry that you guys may know about. And that is AI as workload is not like running Office 365. It’s not like running—I mean, it’s just different. It’s different in so many different ways. But also, when you’re sitting there as the leader in enterprise cloud revenue, it’s not always super obvious that the data center architectures are starting to blow up and grow in a way that creates its own category. And now we’re talking about these AI factories. Making sure we are covering the best of what’s available in the startup world that could be part of those future data centers takes a lot of thinking about what are they doing right now with their business? What do they want to do? And do they really have all the tools, the best tools out in the market to make it possible? And some of those companies—by the way, I’m happy to be a backer of NI Systems founded by Professor Ming here. And I’d say it’s finding this—where is the innovation that they need, and is it in their hand? Can we just invest now with confidence? Because I’m very sure it will be part of those future plans. That’s a judgment call I try to make with every deal. 

YRUEGAS ALMANZA: Yeah, the infrastructure is really interesting. I recently heard about another tech company thinking about data centers in space, even so—

STEWART: Yeah, yeah. 

YRUEGAS ALMANZA: I think that leads well into the next question, which is, I mean, you’ve been around some really interesting technologies. My question is, how do you identify which technologies are truly transformative versus those that are just hype? 

STEWART: Yeah, being technically oriented, being a scientist or an engineer and then being a VC, there’s a double-edged sword to this one is. Yeah, you feel like you could dig into deeper technical, deeper, deeper science, or deeper engineering in a deal. You can also get completely romanced by something being new when there isn’t a strong business behind it. And I think that, one, being fearless to like, OK, this is technically complex. Can I either meet somebody who can tell me more about it? Do research on my own to make these conversations more effective. Or through the syndicate, through syndicating a deal, can I find enough other investor voices that reflect a perception of lower risk than the market might regard this? And it’s a meme right now. This data center is in space. Or what is wrong with the idea of putting a terrestrial data center server up in space? That’s a perfect example of where it’s both simple and complex. There’s dimensions to look at this new technologies that might be needed to make that more effective. Can you go out and read up on that as an investor? Yeah. Can you find other people who are just generally intrigued by the idea? Yeah. But can you sit there and just critically analyze: What does a business like that look like? Can I actually put down: Where are the operational and margin advantages that would come from that? Because you’re otherwise backing a money-losing business. 

YRUEGAS ALMANZA: Building on top of the hyper topic. We did our homework and watched a few of your previous interviews and read some of the articles. And you mentioned that your aha moment for Gen AI wasn’t ChatGPT. It was 2022 using GPT-3 for Jasper and marketing. What was it about that clicked for you? 

STEWART: Yeah, and I talked about this a lot. I don’t how many people catch on to this or if ultimately there’s going to be enough stories written that take this Gen AI moment from ChatGPT forward. I’ll tell you where my intersection with this started. When I first got hired at M12, I wanted to work on something other than semiconductors and electronics and stuff like that. So I got into the metaverse. And one of the things–because I also like video games, I thought, OK, most metaverse games I’ve seen are pretty boring. And you get tired of it quick. One of the ways to make this fun would be to somehow create this bottomless variable but consistent content in a game. And I, along with other people, realized maybe the gaming NPCs would be like—powered by AI would be something because there’s this technology. GPT-3 was in general availability in November 2021. No one did anything. Nothing happened. I mean, it was clearly getting better and better. A lot of people were focusing just on the cost of training. And I think at that time, millions of dollars to train a model just seemed like, how will it ever be paid off? But seeing companies like Jasper grow quickly—and again, I can’t say too many details. But they got to a $85 million run rate in the 2022 pro forma, which is a large multiple over the revenue that the lab whose models they were using was going to generate. Now how is that possible? So this is the first inkling of wait, we’re talking about marking up and distributing, creating a retail phenomenon for AI. That was a gigantic breakthrough. Reselling and today, again, it’s quite obvious because we’re generating tokens and then selling the tokens. They were selling these tokens at a massive markup from what they were getting them out of Azure or other places. So seeing that grow so fast and just out of nowhere and using this technology that no one seemed to appreciate, we had a good eye on it, thanks to our investment in world at that time, because I was already tracking all the Gen AI companies. And I just went around asking Microsoft, ‘What are we doing with this asset, with this Gen AI thing?’ And it’s like, ‘No, this is years away. This is really not the top priority that we have to work on.’ And that’s where, again, the venture funds mandate can come into play. We brought some Gen AI companies to meet the management in the fall of that year. This was about two weeks before the launch of ChatGPT, and we had a tremendous meeting about what people saw. And wait, this is something completely different from what I thought. This is moving much, much faster than I realized. And the signals people, I think, missed were, OK, there’s Jasper, again, helping people write term papers or whatever using AI. There is also this thing in the summertime called DALL-E mini. I don’t if you guys remember this. To me, that was the other, like, no one seemed to pick up on the significance of this. DALL-E mini was free. It was clearly pirate. It was not completely above the board in every way. But they ran up to hundreds of thousands, if not millions, of users. And your Google search traffic for interest in this technology spiked massively. It wasn’t an official product. This was just somebody hacking it together. And you could tell we were not looking for this type of user. We had no inkling that this type of person who’s interested in AI for their personal use existed. 

I think even today as an industry, we are still looking for what is the business case for Gen AI, and the consumer one is almost disregarded. So we flagged this quite early and started investing early and did really well with our first couple of investments. But I quickly could see this is a money game that’s a lot larger than M12. I mean, this is something that Microsoft is going to be deeply involved in. But also, it led me to question a number of times, including today, what are we doing as VCs? Are we catalyzing something new, or are we just tacking on to a phenomenon that’s already very well capitalized? What’s the different strategy that we can bring to it? So still something I wrestle with in my mind. 

YRUEGAS ALMANZA: On that, Michael, what’s your take right now on that internal debate? 

STEWART: Well, I mean, I just wonder are we again—funds can’t change in size at the rate that financings can change in size. So a couple of years ago, and we were talking in the other room norms around what is the seed financing size and valuation and so on. There really shouldn’t be any norms, but there were because of the sizes of the seed funds that were backing these companies. It’s like, there’s just math around how many investments you can really make given the number of partners that you have? How deeply are they involved? Will they be—and what are the odds of a company being super successful versus failing? Well, the norm of a seed round being a few million dollars, like Nvidia’s seed round in 1993 was only $2 million. If you looked at a typical seed round in the middle 2000s, it would also be around a few million dollars. A seed round was announced a few months ago that was $5 billion. And we just had one announced. I don’t know if this is a seed round or not. Another one that’s like $6.2 billion. 

What are we doing here, guys? I mean, so I think, as VCs, we’re going to need to keep going earlier and earlier in the life cycles of the companies—but also be thinking what I’m underwriting must be resistant to or resilient to competition, getting what seems like unlimited money hack. And that’s where, again, I would go back to Microsoft, where I’d say at least I know where there is business for them to get it. If they try hard. We’ll just be seeing companies with—especially in the code generation category, we’ll see many, many companies in the tens of billions of dollars valuation range all capitalized with many billions of dollars in investment. I mean, I would just—at this point, there’s not much I can add to that with our fund. 

YRUEGAS ALMANZA: I want to go back to the metaverse. And you started at M12 focusing on Web3 and gaming. And there seemed to be a lot of hype around it and then complete silence. So I’m really curious how that was for you. And if you can tell us any stories about that time when you were looking at companies, and then when maybe the bell dropped and things changed. 

STEWART: Certainly, one, I’m not a crypto native person. I’m not even a fintech person. I’m a technologist. So the metaverse piece, I felt like, OK, I have some kind of—I have at least a little bit of alpha here because I love video games. I’ll try to play the major titles. I’ll try to play them. There’s too many to cover, but I felt like, at least, that is a business platform in the sense that you’re going to have people paying to be in the metaverse to pay for their time there, whether it’s ad-supported or subscription-supported. And there’s a lot of technologies that can plug into that that show you something like graphical audio content, what have you. And again, my response to that was generative AI. Crypto and Web3 was a little different in that there was enormous pressure it felt like, especially from venture funds that were heavily into crypto assets, to just make a business happen by virtue of the existence of a token. And the investment proposals I looked at—the companies I looked at sometimes lacked boards of directors. They were domiciled in countries where we couldn’t invest. They involved transactions that had no visibility, even though there was a ledger or there’s blockchain to trace back that the transaction happened, and a lot of that just was going to be out of bounds for where we could put Microsoft’s money, where we did successfully find some great places to support the Web3 ecosystem because, again, I’m not a bear on this ever happening. 

I just could see from an investing point of view, you’re used to meeting a company, looking at a cap table, looking at financing docs. Sometimes there is no cap table, there is no ownership records. It was quite loose for me. But a technology company that’s working with Web3, fine. And that’s like space and time. A database of records with zero knowledge proofs that allows you to query the outside world, bring it into crypto-backed transactions or actions. Actually could be an interesting use case for AI even. But the basis of the investment is still a technology company in that case. So I found a couple places to invest there, but I felt I don’t really know exactly what to do about stablecoins being a thing. And we’re going to come back to it. But it still needs to follow something that’s a little bit more regular structure that we can underwrite doesn’t have to be exactly like a Y Combinator, blah, blah, startup, but it could be closer than it is. 

VÁSQUEZ: Now looking into the future. And this is relevant for our Berkeley entrepreneurship community. Similar to the moment you had with Jasper in 2022, what’s that signal that you’re looking for now in 2025? 

STEWART: Oh, I think for sure we have not even scratched the surface in terms of user adoption of AI. I mean, where you guys sit here in Berkeley and in the Bay Area and in the companies you might know around here are far, far ahead of the curve and actually using AI effectively than your median company in the S&P 500. So there’s a long, long way to go as far as people accepting and using AI in their workplace. 

But yeah, I mean, I would be very, very impressed if I saw an AI device that was different from the phone. I’ve attempted to invest in that area. I’ve seen what I think are all of the AI devices coming up that I know of. But I’ll tell you why because at the end of the day, it matters where the AI is generated. It’s going to matter immensely where you experienced AI. And I mean that physically—if it’s close to your face or your body and it’s unobtrusive or it’s part of your everyday habits, you’re going to use it more. You will give the AI back far more process, supervision, or reinforcement than you ever could with a phone. And yet it’s very hard to just visualize what that device is. You guys have seen the Humane pin. There’s a bunch of different cameras like earbuds proposals. There’s a thing you wear around your neck. I would just say that’s the most wide-open thing possible. 

YRUEGAS ALMANZA: Michael, do you like—on that end, do you prefer almost the puck or a wearable when it comes to physical AI? 

STEWART: See, that’s it. I just, I would unconstrain it and just say like, ‘I’ll know it’s fun and interesting to use when I use it.’ It just will be. It’ll just suddenly. And also, is it glasses? Is it AR glasses and stuff? I mean, people would—this whole room you’d all be wearing the glasses if that was the case. You’re not. I mean, it’s been 15 years. I mean, every generation, it gets better and better, but I don’t see a single one of you wearing any wearable AI thing. I mean, maybe watch, but—and yet it’s that close to it being universal. 

YRUEGAS ALMANZA: Well, to close off, we want to go even further more technical, which is something around AI and quantum. How do you personally stay ahead in a field that changes that much? 

STEWART: Well, OK, yeah, so quantum computing is another one of those areas where I think I’ve been through a couple of ups and downs of like, is it over—not hype, but is it over something, or is it underappreciated? And I’ve definitely come back to underappreciated. Full disclosure, I sit on the board of PsiQuantum. Microsoft has their own competing and separate quantum program with a number of partners. And I’m neutrally able to speak to all of them. And I would say, one, what we were thinking of 15 years ago with quantum languages. And I don’t know, some of the early use cases of quantum. I think that part is the one I’ve lost all confidence in because setting up a quantum computation job—and again, I’m a chemist. I’m an inorganic chemist. I can clearly do all these LCAO orbital calculations on my own. If I had a machine to do them, that would be great. HOMO-LUMO gap of some organic conductor or something like that. And yet, the number of people who are going to be able to set up a job like that in their minds, that even mean something, is like next to nobody. It’s going to be very, very hard to just train people like, this is what you could do with a quantum computer, versus the quantum machine acts in conjunction with a discovery platform through AI or others, and can run many quantum jobs without user intervention or a user constructing those jobs. And as we’ll see, these—because PsiQuantum will be online in a couple of years and many others. We’re going to see with real-life proof if that’s true. But it would mean doing something like walking through a couple of thought experiments using AI as a semantic or conversational guide and having exact physically grounded data to complement it nearly instantly. OK, there’s code breaking and things like that, too. But there’s also just these verified rewards you want to do with RL that also, imagine you have a million, multimillion qubit error-free quantum computer to tap, and this is a microsecond calculation. 

That’s why it’s like I would rather, as an investor, if I just take that as a hypothesis, what’s standing in the way to making that connection happen? And I’m already seeing great technologies arise that can do that. 

VÁSQUEZ: I believe we would all benefit from some career advice. So you’ve been both builder and investor. What is it the skill set or the characteristics that make the most successful technical leaders? And what about business leaders nowadays? 

STEWART: Yeah, well, lately I’m just going to be very honest with you. I do wish I could be back as a process engineer right now because there’s just so much cool stuff happening, and you can actually just be a frontline inventor. So any of you who are thinking of a career path and then VC, I think that’s still the best way to go. Get as much experience as you can in the dimension you can be the most creative in. I wouldn’t say exhaust your creativity. You should still you should have plenty of it left. But that could mean areas like scientific discovery or engineering. I came from others that are very, very high alpha. And any of you who are thinking about what to do after graduation go into this AI GTM stuff, like the companies that are learning to use AI and doing it super poorly because they are working with an SI like an integrator who has no, I mean, sorry, has no super deep invested interest in the outcome other than to do an engagement. 

People who are seeing on the ground level, like,  what are the very best tools, what goes wrong when you implement them poorly, and where do you get the highest ROI?. I would work on that as much as possible, and then you’d have an enormously good footing in VC. To go into VC from school or during school, also OK. The reason people will have opinions about, is it good to go early in career or later in career is you do, after a long time, your mind starts flipping back. I’m just used to looking at 10 different options, and I’m down selecting to one. You don’t have the mindset of, “I have this one idea, and what does it take to get it to the top?” Those are your CEOs. The amount of those two dimensions that you work on the broad, being a good gambler—let’s say, a disinterested gambler—and being an inventor are the two areas that you could hone as a future VC. Because you’ll draw on both of those skills if you want to do deep tech. 

VÁSQUEZ: Thank you, so much. We just have one final question, which is a new question we introduced this season for Dean’s Speaker Series. And this is it. So if you had an extra week in your year outside of work, how would you spend it? 

STEWART: OK, you don’t mean an extra week at the end of my life or something like they’re about to move me to the cyborg body and be like, do you want one more week? No. Yeah, if I had an extra week that was work-free. If I could just turn off all the work stuff. This is cliche, but, yeah, I would spend it with my family. It isn’t just because I love them. I love my wife and kids. It’s more because I’ve become aware of the kind of immortality that’s going to be available to us in our lifetimes with regard to what we’re seeing, hearing, and experiencing. So I was never the kind of person to be out photographing every dinner and that kind of thing. I would use the phone to take a picture. Like this is a thing I want you to remember very deeply. And then, it just dawned to me, like, wait, I should just be recording everything all these minutes with the kids because the more you do that—and again, I now think of things like, we are under specking all of the cameras and sensors that go into these recording devices so obscenely. But that’s another thesis to come. Capturing all this data and what we’ll be able to reconstruct from it. 

Let me tell you about the moment in my life that I’m living right now. I have the two kids, two and four years old. My parents are both still alive. I’m able to walk around every day and meet brilliant founders or go to board meetings of my companies that we’re talking about developing AI at the cutting edge. I live in a beautiful area, as you guys, and I’m still reasonably healthy. I mean, there’s a lot about this moment in life that’s like, these are really the best moments. And I would like to be a filmmaker for a week, more in the paradigm of my favorite filmmakers, like Jean-Luc Godard or Fellini or something like that. Go out and just find those moments where I just reconstruct with an internal dialogue, let’s say, these are just the most amazing things I’m seeing and feeling right now. And then, I go back to work. Sometime later I’d like to revisit those and relive those, and I think that’s going to be very, very possible. 

YRUEGAS ALMANZA: That’s amazing. Michael, thank you so much for the time, and I think we’ll move into Q&A. Thank you. 

STEWART: Thank you. 

CHATMAN: Great. I’m going to come over here. 

Stay where you are. I have to use a different microphone today, so thought I’d come and look at you guys, too, for a change. That was so fascinating and interesting. I’m also a Fellini fan, so— 

STEWART: That’s great. 

CHATMAN: Yeah. So there are a couple more questions that go to the issue of what people should be doing now to prepare to be VCs. And I know you started to answer it, but some of these are a little bit more granular. So I thought I’d ask three together. And you figure out how you want to answer that. 

So one is from a full-time MBA student, Martin Lima, who says, ‘What is the one key skill of a successful young VC investor? How does that change from VC to corporate VC?’ The second one is for those interested in working in VC: ‘What skills should we develop during our MBA?’ Similar, and then, ‘When hiring for a role on your team, how much do you value founder operator experience versus previous VC experience?’ 

STEWART: OK, so first one was what skills differentiate financial VC and CVC? Is that right? There are three questions. 

CHATMAN: Yeah. 

STEWART: I mean, I think to be really good at CVC—one of the skills that’s critical for that is business development. Business development partnership. Even though you’re not doing it in the job where I work, it’s understanding if a startup wants to work with Microsoft, or one of Microsoft’s partners, like a huge company, what is that going to look like? Is it just we sign a partnership, we have a little announcement ignite, and then everybody’s happy after that. 

No, I mean, these are actual divisions of who’s making money off of what. It’s good to know a little bit about what those details actually are. And the more that you do of that broadly, the more you get a range of like, this is an unbelievably transformative partnership that changed the destiny of a company to this is a disaster they never should have signed because the gamut is even wider than that. But as a CVC, you’re assessing companies not just in their ability to win. I was going to say category, but just win at all and survive in the venture environment. But if they were going to be a partner, would they be one of a dozen partners of that type or something completely unique? So that skill matters to me. Now where we hire for my team, it’s still you have to be an experienced investor. So I tried at times to do job training at the partner level at our fund. And there’s too much to optimize if you don’t already have shown, it can’t already have shown the ability to internalize that many things because it’s a lot. It’s not only aiming for what is the best deal because we go toe-to-toe with the Sequoia a16z. We compete with the top funds for our deals. Sometimes win, but it’s knowing that. But also it’s Microsoft’s money you’re investing. So don’t go off and invest in something totally irrelevant. I needed to see that in demonstration somewhere else before I’ll give somebody a chance on my team. 

CHATMAN: Great. This may be my very favorite question. It says, ‘Are we in an AI bubble? If so, how is M12 planning? If not, why not?’ 

STEWART: Yeah, yes, we are in a AI bubble. I don’t think that’s too controversial. Should you still be investing? Should you still be believing? I also think yes because I’ve lived through a couple of the technology swings in the Valley. And even though many people may say this and it may be hard to believe, but this time, it really is different. There are so many things about this one that are different, and yet we will definitely see some secular losers. 

I mean, some of the valuations that you underwrite on, remember, you’re going to need more capital. You’re going to have to offer investors a step up. Just think you can put—you can paint yourself in a corner like this. And that’s happened in our industry before three different times. Did it mean that it ended the phenomenon of the internet or other things? No, it didn’t. It just meant everyone’s not going to come out of that unscathed. So as far as the market’s capacity to do more billion-dollar-plus seed rounds, that’s something I just think is—I very much question we can continue that. You have literally Jeff Bezos putting on his gloves again. What kind of team does it take to go to that next level? It’s like literally, what? The CEOs of existing trillion dollar companies going into being founders of startups. What’s the step after that? That’s what I’d ask. 

CHATMAN: Great. Here’s a question about sourcing startups. How do you do it? What percent is inbound versus outbound? 

STEWART: What would outbound be for me? 

CHATMAN: Probably, you finding it rather than it coming to you. 

STEWART: Yeah, if I think about the last couple of deals that I’ve done, half of them have been sourced by our team, our associates, or by me, personally. Just by meeting a company at a pitch event or reading up about them and reaching out. Others could come from other VCs who are raising money for their companies. We have good relationships with many funds, pretty much everybody. So there’s those two ways in are both soft intro ways. They mean someone has already integrated into the VC ecosystem. Because I get this a lot. I hear this a lot, and I’m thinking about it. There are these people. There are the people who are very, very promising entrepreneurs, awesome teams who don’t know any VCs. The process of getting to know an investor can be actually quite difficult to get time to be known before you’re raising money, and that’s a lossy process that, again, besides the effects on diversity, geographical, demographic, or everything that comes into the funnel, we’re just missing out on some great people. Period. But from where I sit, I always ask a few questions as to whether it’s a good M12 fit beyond just is it a good venture fit. Such as what do they think about working with Microsoft? I don’t necessarily put it in those exact words, but to compare and contrast. If I meet a team that’s just like, ‘Oh, I hate Microsoft products and teams. And I would die before using Windows and all the stuff.’ And you do meet founders who are like this. I don’t know if it’s a test to see if I’ll—because I don’t consider it abuse. It’s not my job. I don’t care if you don’t like Windows or not. Versus a founder who’s building something interesting in AI. They either they may have worked at Microsoft. They may just know the world quite well, and they’re like, ‘What I would really love to do with them is this thing. It’s not possible right now, but with help, it could be possible.’ Doesn’t it make sense I would rather work with that second entrepreneur? I mean, there’s a plan that we already are thinking about together. There’s something clear that we can do together. We’re already aligned before they’ve written a check versus a founder who I like. I don’t, he feels it’s funny to not like a big corporation. OK, cool. I mean, maybe I would invest, maybe not, but I would certainly find the chances of us doing something interesting to be higher in the second case. 

CHATMAN: Great. This is an interesting one. Are there examples of founders choosing—well, yeah, so this is exactly on that theme. Examples of founders choosing M12 over traditional VCs. If so, why? And do you know that? Do you know if they have other offers? 

STEWART: Oh, yeah, yeah, and sometimes when people ask this question, they may be asking about a term sheet process. Competitive term sheets for me are like—having won a number of them, I just feel way, way over indexed in how important they are for fund. Unless your job is I need to capture all of the top companies in a category, which is sadly, I mean, not sadly. Just it’s a fact that is the world we’re in right now. For tier 1 VC, it is no longer I’m looking at this area, which is AI applications or AI financial analyst. I pick the one company that’s going to win as the series A. I sit on the board. That becomes my funds bet. Everybody else doing it as an idiot. They’re going to lose. This is the one bet that will do it. 

Today’s world is, they’ll do five companies in the category because missing one could mean the end of the fund. And I’m being serious: labs, applications, infrastructure. So where we come at, it is not, one, we’re a lot less likely to invest in a competitor. Two, I can introduce you to people that if they aren’t direct customers, they work with all the big Fortune 100 companies. 

CHATMAN: Yeah. 

STEWART: And two is, by investing in you and working with us, we will bestow upon you a enterprise ready optics and vision. Again, it’s up to you to not lose that. That is very, very valuable as a company of early stage. So we still, again, as long as we’re protective of how we’re investing. Is it relevant to Microsoft and their companies? Do obey a certain ethical security, whatever standard. That’s how our brand stays strong. 

CHATMAN: OK, I think I have one more that we can have time for. What kinds of questions would you expect to hear from the investors as a founder, and how to build trust with founders as an investor? 

STEWART: For me, again, because I’m technical, nothing is more fun to me than getting into a conversation with a founder about what is really cool about where this technology could go and people who fake that and try to just, I don’t know, hello, fellow kids type of thing their way into those conversations. Usually it’s a pretty polarizing reaction from founders. They’re like, ‘Wait, this guy is bullshitting me.’ And it’s like trying to sway me when it’s not authentic. I also don’t take meetings with founders unless I find the technology to be interesting. So I’m selecting in due to my own bias. But if I find if you can do that kind of ideation, vibe session with the founder, you’re already on the right wavelength to do a deal. You’re already thinking about these are the cool things we could work on together. 

It’s not all going to happen, but technical founders—that’s why from the beginning, I’ve always felt like I can be an investor who’s willing to really hear out what a technical founder is trying to put out there. And it’s not always easy for them. It’s not always—the words aren’t always super smooth. So a little bit of empathy, but also willingness to listen and be in that place together for 20 minutes is crucial to winning over these highly pursued founders. 

CHATMAN: Yeah, great. Sounds like Students Always, Question the Status Quo. We have those people here, just for the record. Michael, thank you so much. Rey, Gonzalo, thank you, for your awesome questions. Congratulations on your incredible success. Thank you, so much, for coming back. I think we all learned a tremendous amount today. Let’s give a warm thank you. That’s great. Fantastic. Thank you.