Portfolio Construction Is Changing. But So Is the Company Inside the Portfolio

500 Global recently published a thoughtful essay on what AI may mean for venture capital. Their argument is one that will feel familiar to anyone who spends time with founders today: if AI changes the economics of building companies, investors will inevitably need to rethink how they construct portfolios. Companies may require less capital, reach meaningful milestones more quickly, and arrive at profitability on a different timeline than previous generations of startups. If that happens, reserve strategies, ownership targets, and deployment models are likely to change as well.

We think they're asking exactly the right question. What struck us, though, was that the conversation seems to begin halfway through the story. Before we rethink portfolio construction, we should probably spend some time rethinking the venture-backed company itself.

For decades, venture capital has operated around a relatively stable model of company building. Founders raised capital in order to hire people. As companies grew, they added engineers, product managers, designers, sales teams, marketers, and managers to coordinate an increasingly complex organization. Growth and organizational complexity tended to move together. The more successful a company became, the larger it became as well.

That model has been remarkably durable. It has survived the internet, mobile computing, cloud infrastructure, and SaaS. Although technologies changed dramatically, the underlying assumptions about how companies were built remained surprisingly consistent.

AI may be different. The reason has less to do with coding assistants than with something more fundamental: the economics of coordination.

Nearly ninety years ago, Ronald Coase argued in The Nature of the Firm that companies exist because coordinating work inside an organization is often more efficient than coordinating through the market. Hiring employees, building departments, and creating management structures weren't arbitrary choices. They were efficient responses to the technological constraints of the time. Information moved slowly. Expertise was scarce. Communication was expensive. Companies evolved as the best available mechanism for organizing increasingly complex work.

The important implication is that the structure of the modern company was never inevitable. It reflected the constraints under which it operated. When those constraints change, organizations tend to change with them.

Over the past two years, we've watched founders move from idea to prototype in days instead of months. Small engineering teams are shipping products that previously required far larger organizations. Designers are exploring dozens of concepts before lunch. Product managers can test ideas long before assembling an entire team around them. Research from GitHub, Microsoft, and others suggests that AI is meaningfully accelerating software development, even if the precise magnitude is still being debated.

It's tempting to think of these as productivity improvements. We think they may represent something deeper.

We've observed this firsthand at Misfit Labs. Building AI-native companies has taught us that the biggest breakthroughs don't come from asking AI to complete individual tasks. They come from redesigning how work moves through the organization. When humans and AI agents share context and structured workflows, coordination becomes cheaper, decisions happen faster, and small teams can execute with the leverage of much larger organizations.

If companies require fewer people to accomplish the same amount of work, the relationship between capital, labor, and growth begins to shift. Capital no longer exists primarily to finance headcount. Organizational size becomes a weaker proxy for capability. The question changes from How many people do we need? to What combination of people, AI, and systems produces the best outcomes?

That is a different organizational problem than the one venture capital has traditionally funded. This is why we believe the conversation about portfolio construction is really a conversation about company construction.

The power law is unlikely to disappear. There will still be exceptional companies that generate exceptional returns, just as there have been through every previous technological transition. We agree with 500 Global on that point.

What may change is why those companies become exceptional in the first place.

Historically, many of venture's biggest winners distinguished themselves by raising more capital, hiring faster, expanding more aggressively, and executing at greater scale than everyone else. Those capabilities are still valuable, but they may no longer be the primary source of advantage.

The next generation of outliers may instead be defined by how quickly they learn, how effectively they orchestrate AI, how intentionally they design organizations, and how rapidly they translate insight into execution. Their competitive advantage may come less from accumulating resources and more from compounding leverage.

That distinction matters because it changes what investors are actually investing in. When we talk about AI-native companies, we often focus on the technology itself. In our view, that misses the larger shift. The defining characteristic of an AI-native company isn't simply that it uses AI. It's that the organization has been designed around a different set of assumptions about how work gets done.

That affects hiring. It affects management. It affects product development. It affects engineering. It affects how many products a company can build simultaneously and how quickly it can move between them. Those aren't isolated changes. They're all consequences of the same underlying reality: AI changes the cost of creating, coordinating, and scaling knowledge work.

Seen through that lens, portfolio construction becomes a downstream question. If a team of fifteen can accomplish what previously required one hundred and fifty people, what exactly should a Series A finance? If AI-native companies become profitable earlier, how should investors think about ownership and reserves? If one organization can launch multiple products from the same underlying infrastructure, is our traditional definition of a startup still the right one?

These are investment questions, certainly. But they're also organizational questions. We suspect that over the next decade, venture capital will continue adapting to AI. It always has adapted to technological change. The more interesting possibility is that the object venture capital has optimized around for the last fifty years may itself be evolving. The portfolio is changing because the company is changing.