Own Your Intelligence: Opening Remarks

Good morning everyone, and thank you all so much for being here. We have about 80 portfolio company founders and AI leaders in the room to explore a very timely topic: owning your intelligence, or sovereign AI. Today's event is meant to be half rallying call and half technical how-to, and so we have stacked the agenda with high-substance technical talks and demos so that we don't just talk the talk of building your own AI, but actually learn how to get there together. Thank you all for taking the time out of your mornings to join us. I know your time is incredibly precious, so let's dive right in.

Let's start with the opportunity. What is sovereign AI? A sovereign is an independent state that has total self-governance, and sovereign AI refers to companies owning their own intelligence without external dependencies, down to the weights.

An important nuance here: we are definitely not telling our companies to get off Opus or GPT. That is definitely not the message. For coding agents, for desktop work, for frontier-level APIs, the closed model APIs are wonderful. But what we've observed is that more and more of our companies are going down the path of wanting to build their own AI capabilities in parts of their products, and vertically integrating toward owning more of this intelligence. Today's session is meant to equip companies that are starting to go down that journey.

We're obviously not alone in this idea. In the last month, the rhetoric around sovereign AI has escalated sharply, with folks like Alex Karp and Satya Nadella speaking up in support of companies owning their own intelligence. Just last week, Jensen Huang led the charge in making sure that open-weight models remain available in the US, and it was awesome to see the near-unanimous wave of support. I think the message is very clear: companies want to own their intelligence. They want to own, not rent, their weights. And I think it's simply because intelligence is too core, too fundamental of a property to just outsource.

We're glad that sovereign AI is in the zeitgeist right now, because we think it's a good thing for the world. On one hand, you have centralized intelligence, where a single, all-powerful AI powers more and more of the world's GDP as a black box, sucking in all the data exhaust, all the data flywheels from the rest of the world. On the other hand, you have decentralized intelligence, where the whole world builds on top of a solid core, but every individual person or company builds their own intelligence on top: bespoke to their own data, their own industries, their own personalization, their own way of working, their own taste. The ecosystem flourishes and individuality triumphs. No single company swallows the rest. I think this is a much more optimistic view of the world.

We work with dozens of companies that are going down the journey of building their own AI. Here are the biggest reasons that we've seen people move.

Reason number one is cost. Especially for low- or negative-margin companies, sovereign AI isn't a nice-to-have, it's a must-have. Ironically, the more successful your AI product is, the higher your AI COGS tend to be. And so it's actually the companies that have been most advanced in their deployment of AI that have been the first to go on this journey of owning their own models.

Reason number two is speed. In certain domains, coding is one of them, security is another, a small distilled custom model can beat a large general one because speed is so important.

Reason number three is performance. And this is a relatively newer one. I would say last year most companies were not choosing to own their intelligence to generate better performance. But we're now at the point where open models can outperform closed ones on your domain, and we're going to spend a lot of today's agenda talking about how to get there.

And then reason number four: controlling your own destiny. Anthropic and OpenAI, I actually think to their credit, have been really wonderful partners to a lot of the ecosystem. But companies are increasingly finding that they want their own set of independent legs to stand on as well.

Does anybody here come from the crypto days, or remember the crypto days? Do you remember this meme? In the crypto days, there was this meme for the DeFi degens: not your keys, not your crypto. If somebody was custodying your crypto for you, it fundamentally wasn't yours. I hereby present the AI version of this meme: not your weights, not your product. For a product to be truly yours, I think it's reasonable to think that you need to be able to control and custody your own weights.

Pat shows this slide at AI Ascent, talking about the race for the application layer. [You look so proud of yourself.] The punch line is that both the AGI labs and the application companies are racing to be the user-facing product from different ends: the foundation model labs from the model side, and the application companies from the user back. I think we're seeing a new dynamic emerge now, which is that the battleground is increasingly becoming not only the race for the application layer, but the race for the intelligence layer. This battleground is no longer only about who gets to control the product, the UI, the go-to-market, the wrapping. It's about who can own the intelligence itself and shape better intelligence in the product. The product is the intelligence, and the newest battleground is for the intelligence layer itself.

And so the hottest new labs, in my opinion, are actually the applied research that we see coming out of companies right now: like Harvey, like Factory, Glean, OpenEvidence, Semgrep, Ramp. The list goes on and on. The research is spanning everything from evals and benchmarks to harness engineering to new algorithmic techniques for fine-tuning, and a lot, lot more. We started this morning talking about centralized versus decentralized intelligence, and I think it's really wonderful to see the amount of innovation that is happening in the democratized intelligence world. I actually think the application companies are the newest neolabs.

Okay. I assume that everyone here today is pretty bought into this journey. Let's assume that you want to build your own lab, build your own models. How do you go from zero to one to 100? We're going to do something a little bit different today. I'm going to lay out an opinionated framework and technical roadmap. Take that with a giant grain of salt: every company is different, and I'm not technical. But I hope it provides a useful starting point for how to think about building your own intelligence.

Step one to owning your intelligence is strategy: what parts of your AI do you want to own, and what parts do you want to rent? Step two is team: figuring out how to staff and organize people toward the production of intelligence. Step three is legibility. I really think this gets glossed over and is incredibly important, so more on this later. And then finally, step four, we're going to talk about a technical roadmap: what are the building blocks you need to assemble in order to build your own intelligence?

So let's dig in. Step one: defining which capabilities you want to own versus rent. Sovereign AI isn't binary. You're not 0% or 100% sovereign. An important part of this strategy is to draw the lines for which intelligence you want to own and which you're comfortable outsourcing. Here's a useful framework to think about what parts you want to own versus rent. I think there are four important factors that go into this. One is cost: how important is this cost line item to your overall COGS? Two is speed and latency: is it a P0 or not? Factor three is performance, and this is where it gets very interesting. It used to be that you would choose open weights when you didn't care about performance. Now we're getting to the point where, in certain domains, you may be able to get better performance by tuning models on your own data. And then finally, proprietary data: are you in a domain where the data you're giving the model to improve it is super proprietary to your business, or less so?

Here are some examples of how companies have decided to make this trade-off. In coding, you have both agents and autocomplete. On the agent side, these are still mostly rented today, because you want strong out-of-the-box performance and latency isn't a P0. On the other hand, for tab autocomplete models in coding, you really, really care about speed, and these API calls are so frequent that the costs really rack up. And so most tab autocomplete models now run on sovereign intelligence. [We work with a cybersecurity company in stealth that owns its models primarily for speed and performance and the ability to post-train the model in very bespoke ways.] Bio companies are moving to their own models because of the value of proprietary data in that space. So I think this is just a useful framework: which AI capabilities do I want to own versus rent?

Step two: assemble a team. I've shown two profiles of labs leaders here just as examples. [Niko comes more from the research side of the house, having done research at Apple and then at Google Brain. Alex comes more from the engineering side, having held multiple engineering roles at Microsoft and then Ramp.] And I show this just to say there are multiple paths to nirvana. Depending on the flavor of research you're going to be doing at your company, whether it's going to be more fundamental or applied, there are different profiles that work for a labs leader.

I also think it's important to think about how to design your organization. Traditionally, AI teams have frequently been organized hub-and-spoke: you have a single platform team supporting different application product teams. What I've seen is that a lot of companies are shoehorning this AI platform team into doing the sovereign AI stuff as well. I'd encourage folks not to do this. I encourage people to start from scratch here, because this fundamentally is not a platform services capability. You want people who are able to think on their feet, think on the frontier, and produce frontier-level research. It's such a different flavor of research, and you want them to be playing offense, not just servicing teams. We've seen small de novo teams get very far here. Harvey, for example, has published a ton of research [with just a team of seven]. So start small, start from scratch, and consider making it your own lab.

Step three: legibility. I think this is totally underestimated, because my guess is a lot of people in this room are doing wonderful research in-house, and not all of it is very externally legible. Winston Weinberg talks about how the responsibility of a CEO is twofold: one, drive substantive results, but two, control the narrative, control legibility around what you're building. And I totally agree. When it comes to owning your AI stack, legibility really matters, because every single buyer right now is choosing their AI champion. They're getting the same pitch over and over again, and they're trying to discern which vendor is sophisticated enough to take them to the promised land. They want to choose people who know what they're doing. Increasingly, that means putting out your own research. Being legible here means doing excellent technical marketing, maybe having your own separate branded labs or research group, publishing research with high taste. All of this matters a lot, and I think it goes overlooked. For all the Sequoia companies in this room, I would really, really push on us to think about this.

And then finally, step four: setting your technical roadmap. At a high level, here is the rough journey that I see companies take, and every company goes on a different journey. First you set your strategy. Second, defining evals. This is so important. It is unglamorous work, it is not fun work, but the more that you do upfront, the better positioned you are for everything after. I think this is really crucial to get right at the beginning. Next, we see companies starting to play with model routers, with harnesses. Some companies find they can get good performance with out-of-the-box models. Others are finding strong performance gains from post-training, in some rarer cases needing to move into mid-training or pre-training. And then finally, setting that machine up so that live customer data is actually creating a feedback loop, where your intelligence is improving with every customer interaction. This is the rough journey that I see people go on. Again, every company is very, very different, and I'd encourage everyone in the audience today to compare notes with the people around you. Everyone is somewhere on this journey.

The beauty of owning your stack is that you can actually drive frontier-level performance now. This is somewhat new, and in large part this is thanks to the newest open-weight models, especially Kimi K3 and GLM 5.2, being extremely good. Because the weights are available, they're actually much more malleable than working with the closed APIs. So you start with a baseline that's already close to frontier, and then with a good enough technical roadmap — with strong post-training, prompt and harness engineering, and online learning — you can actually reach better-than-frontier performance by owning your stack. This is new for 2026, and I think this is very, very important. This is a big part of why people are starting to think about owning their intelligence.

I like diagrams, and so in an attempt to orient us all, here is how I think about the stack from an infrastructure perspective. On the left-hand side is production. This is your user-facing intelligence, the stack that drives every token your user ends up seeing. Fundamentally, I think of the production stack as a harness on top of a model. Alongside that, you have a development stack: the tools and vendors you use to get your intelligence good.

In the closed model ecosystem, this entire stack is very simple. You have foundation models like Opus and GPT, and you have the harnesses that come out of the box with each. You can get quite far with this stack, including building your own harnesses and prompts, feeding context into the models, and doing your own evals. But you're not really having to collect a ton of data, and you're not really having to train your own models, so it's a much simpler stack. I'd say this is a higher floor, but a lower ceiling, because you don't have the ability to take your own online production data and improve your own intelligence.

The minute that you start to think about owning your own intelligence, it is like opening a Pandora's box, because that beautiful, clean API call now becomes training your own models. Instead of having a single model API and good performance out of the box, you have to choose an open source base and do a lot of post-training on top. For your harness, you've got a choice of several open harnesses, and then it's up to you to configure the harness, the logic, the tools, and the context, versus taking an agent that just works out of the box.

An aside here: context is really, really important to driving performance, and there are several different flavors of context driving these models: a vector database like turbopuffer, an enterprise knowledge graph like Glean, open source connectors via MCP, and even novel approaches to context. [Dan Biderman from Engram is here. They're doing novel research around encoding context in the weights themselves, so I'd encourage anyone who wants to chat about memory to go find Dan during one of the breaks.]

Alongside the production stack, the development stack becomes way more important when you own your own AI. You have to carefully monitor evals to know how your intelligence measures up, and watch how the model performance is drifting in production. You need a lot of high-quality data to train the models for your domain: sometimes expert trajectories, sometimes synthetic data, sometimes RL environments. And then finally, you need to think about how to set up online learning for your system, so that your intelligence gets better and better with every single user interaction.

And so the way that we've set up today: we've picked a series of technical workshops to give you deep dives into everything you need to build your own AI, other than pre-training your own models. Lin from Fireworks is going to lead a workshop on post-training. Harrison from LangChain is going to lead a workshop on harnesses and evals. Brendan from Mercor is going to lead a workshop on RL environments, synthetic data and more. Trajectory is going to lead a workshop on online learning. And then to bring it all together, Harvey is going to lead a workshop on how they approach building their entire AI stack and strategy — just yesterday they announced Harvey Research, and you'll see that many of their technical partners are actually speaking today. We've really gone all out to get the best possible lineup of speakers today, both inside and outside the portfolio.