Episode #92 - Three Breakthroughs: A Survey of the AI Landscape

Frequently Asked Questions

FAQ

• Where is the transcript for this webinar?

  Sam:
  Welcome to Tech Optimist, where we don’t just talk about the future—we break it down, analyze it, and watch it unfold in real time. I’m Sam, a producer and host of this show. This week, we’ve got three massive breakthroughs shaping up the AI world:

  • A Chinese startup builds a model that rivals OpenAI on a shoestring budget.

  • The US government bets big on AI supercomputing with a half-trillion-dollar initiative.

  • OpenAI launches Operator, the first AI assistant that doesn’t just respond—it takes action.

  The stakes are high, the innovation is faster than ever, and the competition is getting fierce. So buckle up—this episode is all about power, progress, and the next wave of AI dominance.

  Speaker 2:
  Do you have a venture capital portfolio of cutting-edge startups? Without one, you could be missing out on enormous value creation and a more diversified personal portfolio. Alumni Ventures, ranked a top-20 VC firm by CB Insights, is the leading VC firm for individual investors. Believe in investing in innovation? Visit av.vc/foundation to get started.

  Sam:
  As a reminder, the Tech Optimist podcast is for informational purposes only. It is not personalized advice and it’s not an offer to buy or sell securities. For additional important details, please see the text description accompanying this episode.

  If you thought last week was a big one for AI, buckle up—because this week just changed the game.

  Mike:
  Hi, welcome to Alumni Ventures and our Three Breakthroughs discussion. I’m here again with Lucas this week. Wow—I mean, you always think, “Oh, it’s been a busy week in technology and news,” and yeah, a lot is going on.

  Sam:
  All right—a Chinese AI startup just made the entire industry do a double take with DeepSeek, a model that’s proving you don’t need billions of dollars or an army of GPUs to build something groundbreaking. They built a model that rivals OpenAI’s best with a fraction of the computing power and budget.

  But was it innovation or imitation? OpenAI has accused them of stealing GPT’s intelligence, sparking a major controversy—which we’re going to get into today.

  Then we turn to Project Stargate, a $50 billion initiative that’s making waves in Washington. This is the US government’s biggest bet yet on AI, aiming to build out computing infrastructure at an unprecedented scale. But does this signal a new AI arms race—and what does that even mean?

  Finally, we’re talking about OpenAI’s Operator. We touched on this last week, but it’s the first AI assistant that takes action instead of just answering questions. Imagine a future where AI books your flights, orders your groceries, or even manages your schedule all on its own. It’s not perfect yet, but it’s the clearest sign of where AI is headed.

  So these aren’t just cool tech stories—they’re shaping the future of global competition, infrastructure, and automation. Let’s jump in and unpack what it all means.

  Mike:
  So we’ve got DeepSeek, Stargate, Operator… I think this may be all AI, but let’s start with DeepSeek. The world had kind of a stroke when this little Chinese company came out with a model people are very impressed by, and they’re giving it away to open source, built in a new and better way.

  I’ve got my points of view. Lucas, what are your bullet points on it?

  Lucas:
  Yeah, just to go back from the top—it has not been a slow news week, but this definitely takes the crown. Everybody seems to have lost their dang minds over this thing.

  I think the most intriguing aspect is DeepSeek’s model efficiency. Unlike the Western AI labs that spent hundreds of millions of dollars to train their models on massive GPU clusters, DeepSeek managed to achieve competitive results using only a couple thousand GPUs over a two-month period on a budget of $6 million—that’s what they’re saying in their marketing materials.

  So it raises critical questions: Have we overestimated the cost of achieving state-of-the-art AI? And if so, what does that mean for the future of AI investment and development?

  Mike:
  Yeah, listen—you can go a million different directions with this. Just to bat the tennis ball back a bit, this definitely points to the fast-follower business strategy, which, with the long arc of venture capital, I’ve seen to be a pretty good one.

  I mean, Apple has usually not been number one into the market. It’s a lot easier to be a fast follower. So I think that’s one thing.

  I personally think it’s naïve to be surprised that China is going to compete in this space—that innovators and entrepreneurs won’t deal with constraints and find new and better ways.

  And I think competition is good. The fact that we’ve got Google, Meta, Elon, and other players pushing each other to make AI better—I think that’s fantastic.

  As a venture capitalist, we weren’t interested in taking on Google in LLMs. I think you want to invest in businesses that benefit from this competition—those who own a customer, own a dataset, own a niche, know a problem better than anyone else.

  Let these guys be in a multi-billion-dollar bloodbath. It’s going to help my business. And speaking as Alumni Ventures as a business, we’re going to use the best models to solve the problems we want to solve. Period.

  Bring the competition on—that’s my point of view.

  And my last thought, before I let you bat it back, Lucas…

  Lucas:
  I’m excited to volley the ball back over.

  Mike:
  Yeah—this is also one of those things in the human experience: “Oh, when we have this much energy, we’ll have enough. When we have this much computer, this will be enough. All the great ideas have happened. It’s too late now. I only wish I’d been around 10 years ago in venture capital and innovation.”

  But I’ve just seen this song over and over again.

  Take Zuck after DeepSeek—he’s announcing a facility half the size of Manhattan. Could these things be more efficient? Can you do more with less? Great. That’s just going to add to what these huge investments will be able to deliver to the world.

  So… your stroke.

  Lucas:
  I think those points are all very valid. Competition will always spur more innovation to the benefit of enterprises and consumers. That’s what we believe as tech optimists. But I do think that there are a lot more stories coming out today, and DeepSeek—at least as of a few hours ago—I don’t think they’ve actually officially responded to this.

  There’s a lot of talk about the use of distillation, which in AI moves away from fair competition toward questions around whether we’re ripping off another company’s IP.

  Distillation in AI is a technique where a smaller, more efficient model is trained to replicate the behavior of a larger, more advanced model.

  Mike:
  It doesn’t exist unless you have done the work on the big model, right?

  Lucas:
  Exactly. It’s like a junior chef learning from a master, and rather than spending years experimenting with every ingredient and technique, the apprentice just closely observes the master’s process, makes those same dishes, and then opens a new restaurant—if I were to simplify the situation.

  And you might say, “Okay, what do I care about as a customer? If DeepSeek did that and now I get the same model for cheaper, that’s to my benefit.” But we need to see the repercussions of this.

  OpenAI is now accusing DeepSeek of really crossing ethical and legal boundaries by using distillation in an unauthorized way. Essentially, they’re accusing DeepSeek of scraping ChatGPT’s responses to then train its own model.

  Mike:
  By the way, I think that’s exactly what they did.

  Lucas:
  Yeah, and it’s right there in all of the user agreement language that you cannot do that. I don’t think we should be surprised that a Chinese-owned company was perhaps engaging in that type of behavior.

  But we need to see where this goes, and it raises really interesting questions for Project Stargate. I don’t know if you want to… I can volley the ball.

  Mike:
  Let me respond to that—which is, yep. This is, again, we’ve talked a little bit about doing hard things.

  The simple example of hardware and software: both are hard. The ability to shortcut software development versus what Apple has done over decades—where they’ve worked the entire stack, layer by layer, harder than anyone—that’s a massive difference.

  This is classic Clayton Christensen, right? About the value chain and the value stack, and how people can take one particular slice and commoditize the business. That is exactly what has happened here, while the race is to vertically integrate.

  Vertical integration can give you a competitive advantage—which is really hard to do and takes a really long time—but if you can achieve that, somebody’s not going to rip you off in the same way that it seemingly has happened here.

  And listen, these are big players with deep pockets who are super smart. It’s the same reason Sam’s getting into robotics and wants to get into chips. They all know that this is the playbook.

  So for me, there’s a little bit of, “Gosh, I can’t believe how surprised everybody is.” Maybe that’s just because we live in this world every day and I’ve seen this playbook for 30 years.

  For me, it was “of course,” whereas the average retail investor who bought NVIDIA stock and sees a correction—their perspective may be a little different. I’ll leave it at that.

  Lucas:
  Yeah, that’s such a good point. You’re kind of getting at the heart of it—even if OpenAI was not open-sourcing their model, they knew there’s still a long road ahead for building large, sustainable moats.

  That’s something you just elaborated on that, on the investing team, we wholeheartedly believe in.

  We wrote a check into a company called Opal that OpenAI invested a lot of money into. They’re trying to use this company to be the hardware interface for interacting with the model when it’s not on mobile or desktop.

  What does that hardware interface look like? They know this needs to be a full hardware and software stack, integrated, to form a more defensible moat.

  Sam:
  So let’s break this controversy down a bit more. DeepSeek recently launched an AI model that performs at a level comparable to ChatGPT.

  But here’s the kicker—they reportedly developed it for just $6 million, a fraction of what US AI firms are spending. This has left many wondering: How did they do that?

  OpenAI claims they have the answer: model distillation—a process where an AI is trained by mimicking responses from an existing model. While this is widely used in AI development, it violates OpenAI’s terms of service if done using their proprietary data.

  US officials have weighed in, with some suggesting DeepSeek may have unfairly leveraged US technology to create a dirt-cheap AI model.

  But this controversy isn’t just about ethics—it’s also shaking up the tech industry and the stock market. After DeepSeek’s model went public, major AI stocks including NVIDIA, Microsoft, and Alphabet saw dips, signaling concerns over cheaper, more competitive AI alternatives.

  Then there’s the issue of content control. Tests show that DeepSeek handles politically sensitive topics differently from ChatGPT, particularly when it comes to China, human rights, and global conflicts. Some argue this highlights the risks of AI bias and censorship, especially as open-source and closed-source AI models continue to evolve.

  So what does this mean for the future of AI? With the rise of faster, cheaper competitors, will AI development shift toward cost efficiency rather than raw power? And if companies like OpenAI struggle to protect their intellectual property, what does that mean for innovation? These questions aren’t just theoretical—they’re shaping the next phase of this AI competition.

  Lucas:
  But I’m really interested in what happens next with Project Stargate.

  Mike:
  Yeah, let’s talk about that a little bit. Again, for our listeners, this was the big announcement in the Oval Office a couple of days or three days after the inauguration, which was a bit of a cluster. And then there was some back and forth about the number that got used and how real that number is.

  Headline numbers are headline numbers—they’re meant to be attention-grabbing. I would discount the details a lot. You can break it apart: what’s year one, year two, year three? How much of this is going to be done using deep-pocketed venture debt? We have big players like BlackRock who have basically said, “Hey, we’re here to loan money against these kinds of things.” So, how much actual equity is getting put forward? It’s not anywhere near $500 billion in year one, for sure.

  This is big-picture stuff. What this is to me is just the pattern, which is big money. A lot of groups are building these enormous GPU clusters, and there’s game theory involved. You can listen to Zuckerberg being interviewed by Rogan about it. There’s a prisoner’s dilemma here—they all know this is a game they have to play, that they want to play, and there’s some non-trivial possibility that this is the last big game we are all going to play in this dimension.

  When software becomes smarter than humans and can do things better than humans—that’s never happened at scale in human history. So all bets are off. They’re all betting the firm. It’s very fun and interesting to watch, and this won’t be the last big cluster to be built.

  I tend to be in the camp that says: for this kind of stuff, computers are everything. Take it from an old man who has the original Mac sitting over here with 128K in it. They came out with 512K and people thought, “Who would ever need 512K in a computer?” Humans are wired for more—we will find uses for more, we will always reach for more and bigger.

  If you build a computer, we will use it. If you’re able to produce energy, it will be consumed. That’s where my point of view is on this stuff. Stargate is big. I think it’s interesting that the government is in the game, at least showing support.

  Related to this is a dialogue about bottlenecks. Clearly, we have issues with chips, memory, and energy. Energy is a stickier one because we’ve had 40 years of stagnation there.

  An unintended consequence of DeepSeek is a wake-up call: we don’t own this market forever as a birthright. If we’re going to compete, we have to continue to innovate. We can’t be hamstrung by regulation that prevents us from building the energy we need to power these systems.

  The choice is simple: we innovate, or we lose. And if you lose in this game, you’re playing a very dangerous game.

  Sam:
  All right, we’re going to pause the conversation here and hop into an ad. Don’t go anywhere.

  Speaker 5:
  Hey everyone, just taking a quick break so I can tell you about the AI fund from Alumni Ventures. Alumni Ventures is one of the only VC firms focused on making venture capital accessible to individual investors like you. In fact, Alumni Ventures is one of the most active and highly rated VCs in the US, and we co-invest alongside renowned lead investors.

  With our AI fund, you’ll have the opportunity to invest in a portfolio built entirely around advancements in AI. This fund consists of 15 to 20 investments in multiple fields where AI is making a huge impact, including areas such as machine learning, healthcare, education, transportation, and more.

  To get started, visit us at av.vc/funds/aifund. Now, back to the show.

  Lucas:
  I think all of that is right. I think this is an investment in brute-force scaling. Stargate represents a bet that scaling up AI hardware to an extreme degree will unlock new levels of intelligence, pushing OpenAI closer and closer to its vision—and not just OpenAI, but also Microsoft as its lead backer, Oracle, the US government, and other players involved.

  This brings us closer to a vision of machines that can reason, plan, and learn like humans. And that has been OpenAI’s stated—

  Mike:
  Yeah. Let me put a little refinement on this discussion too, which is I think there’s a perception—again, kind of related to DeepSeek—that all of these data centers are about training data and getting to the next level of capabilities. But Jensen has pointed out there’s also the inference side of this equation—the actual chewing of problems and solving of things. Once you’ve trained the model, it’s about doing stuff.

  Jensen said that market is a billion times bigger than the training market—a billion. My view is these data centers can’t be built fast enough. They will be put to use. It may not all be for training. These things are compute centers—they’re horsepower. And that horsepower can be applied to a lot of different things fairly flexibly over time.

  So, saying we’re building Stargate to do “X” is a very narrow, short-term perspective on why these data centers are being built. We’re going to need data centers to do old-fashioned stuff. We’re going to need them for training, for learning, for reasoning, for optimization—and then we’re going to need them to actually do things.

  That brings us to the third big piece of news this week, which was a little overshadowed: what I think is the first consumer-grade agent—Operator from OpenAI. I’ve played with it. It’s slow, it’s janky, but it’s a window into the future. There’s a bit of magic there.

  You can see where this is going. Imagine saying: “I’m going to Napa, I’ve got two couples, here are the dates, here are the restaurants I want in order—go to town.” And then I just go back to my day job. Meanwhile, out of the corner of my eye, I watch it logging in, reviewing dates, reviewing screens, occasionally prompting me like: “Hey, no time available—can you go later?” or “We have one—do you want me to do it or do it yourself?”

  If it can do that—and it’s just silently clicking, swiping, filling in dates—that’s huge. Right now, it’s like McPaint in 1987: rough around the edges but you can see the exciting future. With DeepSeek, Stargate, and all the noise in the environment, this is a big deal.

  When computers can start acting with us to do things—a lot of our world, personally and professionally, is basically sitting on computers moving stuff around. If we have something smarter than us helping with that, the implications for society and the world are enormous.

  It’s not there yet, but if you look online you’ll see people experimenting with it. And you can bet that DeepSeek or someone else will soon release their own version of Operator.

  Lucas:
  Yeah, absolutely.

  Mike:
  There’s totally going to be a DeepSeek version of Operator. Now, people have to decide: do I want my data and my credit card information with DeepSeek making reservations for me? Personally, I’m pretty happy and fully occupied keeping up with what’s going on at Gemini, Perplexity, Claude, and others. I’m not rushing to spend time on DeepSeek right now.

  I know what I’d get: a free, almost-as-good version from a company that’s essentially cloning. And I’m very skeptical. When in doubt, I’m not sharing a lot of private information with an entity that operates like DeepSeek does. That’s my personal choice—knock yourself out if you want to.

  Lucas:
  I think that’s largely considered a best practice for tech hygiene. For Operator, we’ve talked about this a couple of times already: OpenAI just continues to be at the front of the pack for bringing top-notch AI consumer products to market.

  As people focus a lot on the enterprise side of things, I’m excited about more enterprise applications. But OpenAI’s strategy has always been to demonstrate how this works for consumers—

  Mike:
  Kind of like Apple.

  Lucas:
  Exactly, as a gateway to a bigger market.

  Mike:
  They don’t always invent it first—you could argue there are ways to do some of these Operator tasks with Claude—but they make it simpler and more accessible.

  Lucas:
  And even in OpenAI’s release strategy, it shows. Their product marketing team knows this is their approach. In the release video, they showed how a user could shop on Instacart using Operator: “Here’s what I want to cook tonight. Here’s what I already have in my fridge. Go load up a cart for me.” That’s the ultimate consumer application.

  You mentioned travel earlier. For about a year now, I’ve been talking with my team about how since ChatGPT’s first release, people have been excited to have an AI travel agent. But no one’s really done it well yet. This is the first real foray into it.

  I’m also relieved we haven’t placed a bet in that space because it looks like OpenAI is going to win that one.

  Mike:
  Yeah, it seems like Operator-like tools are just going to handle tasks for you. That raises an interesting question: do you need a specialized tool, or do you ultimately want to be an airline—or any business—that’s simply compatible with AI?

  I noticed when I was looking at flights with Operator, certain sites seemed more AI-friendly than others. It reminded me of how the world evolved to work with Google. If you had a website, you had to think about SEO and how to make it work well for both humans and search engines.

  I think there’s a similar analogy here: if you’re solving a problem in the world, you’d better be thinking about how to make your product or service easy for AI to work with. Now, AI is smart and will eventually figure out your site or company anyway, but there’s got to be an advantage in making it simpler for agents to interact with you. And frankly, making it easier for AI to work with you probably also makes it easier for a human to work with you.

  Lucas:
  Right. There’s another dimension I like when thinking about the Operator release. A couple months ago, I read an opinion piece from a more artistic perspective. For us on the investing team, it’s obvious—OpenAI has increased our productivity tremendously.

  But the writer said something insightful: “I don’t want AI to help me read and write so that I have more time to do dishes and other annoying chores. I want AI to do the annoying chores so I have more time to read and write.”

  I think that’s a really interesting consumer perspective. With the release of Operator, we’re finally starting to see a world where you can offload tedious tasks and truly get time back—not just for productive work, which is where you and I focus, but also for things in life that you love: more time with family, reading and writing, or just going out and enjoying the world.

  I think we lose that thread sometimes when discussing AI—not you and I, but in general conversations about it.

  Mike:
  That’s a very insightful point, Lucas. It’s important to remind our listeners that this is really the purpose of these tools. But you have to take control of that outcome, because these tools can just as easily have the opposite effect.

  A good example is Zoom. It’s a productivity tool—it lets us have a meeting without flying to Chicago for a two-hour conversation. That’s supposed to free up time. But did people actually use that extra time for improving their health, relationships, or creative pursuits? Or did they just end up working more hours?

  If that’s your goal, fine—but make it intentional. Humans often don’t do that. We get sucked into the flow of work, and suddenly, we’re not where we wanted to be.

  With AI, you’ll have a massive leverage point to get more of what you truly value—or not. You have to be aware and conscious about it.

  For example, if you value reading, AI can absolutely help you read more and read better. It can summarize books, analyze what you’ve read, suggest new material, and even hold conversations with you about it. Imagine having a 24/7, always-ready, personalized book club.

  That’s how you can use these powerful technologies to align with your personal goals.

  Another great conversation, Lucas. Hopefully our community got a point or two out of it, and we’ll do this again next week.

  Lucas:
  Thanks for having me—

  Mike:
  Maybe next week will be a slower one.

  Lucas:
  Yeah, maybe we’ll have one story that isn’t directly about OpenAI.

  Mike:
  Yeah. Okay. Be well.

  Lucas:
  Take care.

  Sam:
  Thanks again for tuning into The Tech Optimist. If you enjoyed this episode, we’d really appreciate it if you’d give us a rating on whichever podcast app you’re using. And remember to subscribe to keep up with each episode.

  The Tech Optimist welcomes any questions, comments, or segment suggestions. Please email us at info@techoptimist.vc with any of those. And be sure to visit our website at av.vc.

  As always, keep building.