The Network Is the Fund

Venture Capital Was Built for an Economy That No Longer Exists. What Comes Next Will Be Better.

Venture capital is a $700 billion asset class in the U.S. It should be $4 trillion. The gap between those two numbers explains more about American economic stagnation than any policy debate, trade war, or interest rate cycle. It also explains why we are about to see the most productive decade in a century.

Here is the situation. In Q4 2025, just 8% of VC deals accounted for 75% of all venture dollars deployed. Deal counts for transactions under $100 million hit their lowest level since 2012. Meanwhile, 117 mega-deals totaled $56 billion in a single quarter. The money is concentrating, not spreading. AI captured more than half of all invested VC dollars globally in the first half of 2025. Almost every other sector saw decline or stagnation.

This is not a market correction. It is a structural failure of an innovation financing model that was designed for a world of centralized knowledge, hierarchical corporations, and geographic concentration.

That world is ending.

The Sloan Inheritance

To understand why venture capital is breaking, you have to understand what it was built on.

In the early 1900s, American corporations were chaos. Alfred Sloan stepped into General Motors and created the modern organizational structure. Divisions. Reporting lines. Professional management separating ownership from operations. The model worked so well that it became the template for every large enterprise in the country, and eventually the world.

Sloan’s innovation solved a real problem: how do you coordinate thousands of people making millions of decisions when no single person has enough information to make them all? His answer was hierarchy. Push information up. Push decisions down. Standardize the interfaces.

That organizational architecture created a specific kind of economy. Large firms dominated because they could process more information internally than small firms could access externally. Scale meant knowledge advantage. Knowledge advantage meant pricing power. Pricing power meant profit margins that attracted capital.

Venture capital grew up inside this economy. Its entire model assumes you are funding an entrepreneur who will build a company that looks, eventually, like a Sloan-era corporation. Raise a seed round. Hire. Raise a Series A. Hire more. Build the org chart. Scale the hierarchy. Exit to a public market or a larger hierarchy that wants to acquire your product.

Every assumption in that chain depends on information being expensive to acquire, slow to move, and concentrated in specific geographies and institutions.

Those assumptions are now false.

The Coordination Problem, Solved Differently

In 1945, Friedrich Hayek published an essay that venture capitalists should have tattooed on their forearms. “The Use of Knowledge in Society” argued that the economic problem is not how to allocate known resources. It is how to use knowledge that is never available in concentrated form. It exists only as “dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess.”

Hayek’s answer was the price system. Prices encode information about supply, demand, and scarcity that no central planner could ever aggregate. The price system works not because anyone designed it, but because it lets millions of people act on local knowledge without needing to understand the global picture.

For 80 years, this insight was an elegant theory with limited practical application beyond commodity markets. Corporations still relied on internal hierarchies to coordinate. Venture capital still relied on geographic proximity to Silicon Valley, personal networks, and partner meetings in Menlo Park conference rooms.

Then AI started doing something interesting. Not replacing workers. Not generating cat memes. Something far more consequential.

AI began collapsing the cost of coordination.

The Real AI Story

The productivity debate around AI is framed wrong. Economists at Wharton estimate AI will add 0.1 to 0.2 percentage points to annual total factor productivity growth, peaking in the early 2030s. The Kansas City Fed found that productivity gains since 2022 are concentrated in a small number of industries, mostly information services and professional activities.

These numbers are accurate and completely miss the point.

They measure AI the way you would measure a new machine on a factory floor. How many widgets per hour? What is the labor cost savings? This framing captures the first-order effect: AI automates tasks and reduces costs within existing organizational structures.

The second-order effect is what matters. AI does not just make existing organizations more efficient. It makes the organizations themselves less necessary.

Consider what a startup actually needs. A founder identifies a customer need. She needs capital to build a product. She needs talent to execute. She needs distribution to reach customers. She needs information about what customers actually want so she can iterate.

Every one of those needs historically required institutional intermediaries. Banks for capital. Universities and corporate talent pools for people. Retail chains or enterprise sales teams for distribution. Market research firms for customer intelligence.

AI collapses those intermediation costs. Not by 10% or 25%. By orders of magnitude.

A founder in Topeka can now use AI to conduct market analysis that would have required a McKinsey engagement. She can use AI-assisted code generation to build an MVP that would have needed a five-person engineering team. She can use AI-driven marketing tools to test customer acquisition channels that would have required a growth marketing hire. She can use AI agents to manage supplier negotiations, compliance filings, and financial modeling.

The startup does not need to become a Sloan-era hierarchy to scale. It can stay small, networked, and lethal.

The Venture Model Inverts

Here is where the structural break happens.

Traditional venture capital makes money by identifying which startup will become the next hierarchy. You invest in 30 companies. Two or three become large enough to IPO or get acquired by an even larger hierarchy. The returns from those two or three must cover the losses on the other 27.

This model requires massive companies as the endpoint. It requires billion-dollar exits. It requires the kind of winner-take-all dynamics that only emerge when scale confers unassailable advantage.

But when AI makes it possible to serve customers without building a 500-person organization, the exit math changes. You do not need a unicorn. You need a portfolio of profitable, capital-efficient companies generating cash flow in markets that traditional venture capital ignores.

Think about what this means for American innovation. Right now, total factor productivity growth in tech has been exponential. Every other industry code in the country is flat. Healthcare, agriculture, construction, education, manufacturing, logistics. These sectors employ most Americans and produce most of GDP. They are starving for the innovation capital that has been hoarded in San Francisco.

The median VC fund size outside California, New York, and Massachusetts is $10 million. Inside those states, it is more than double. If you are building an agricultural technology company, your talent and customers are in St. Louis, not Silicon Valley. If you are building a food-as-health platform, your customers are the 75% of American adults who are overweight or obese, and they are everywhere.

The network-centric model does not require entrepreneurs to relocate to Sand Hill Road. It requires capital, knowledge, and customer signals to flow freely through networks that operate regardless of geography.

Information Arbitrage at Scale

In the mid-1990s, when every bank in America was running away from technology lending, a firm called Hambrecht and Quist doubled down. H&Q sat in the middle of every IPO, never leading left but always present. They were not prescient. They were positioned. Their presence in the middle of the startup ecosystem exposed so much information that investment arbitrage was almost inevitable.

H&Q understood something that Shannon’s information theory makes precise. The value of a signal increases as the noise around it decreases. And the best way to reduce noise is not better analysis. It is better positioning within the network where signals originate.

Venture capital has always been an information business masquerading as a capital business. The best investors do not have better analytical frameworks. They have better signal access. They hear about the deal before it is a deal. They understand the customer pain before the entrepreneur does, because they are embedded in the network where that pain expresses itself.

What happens when AI democratizes that signal access?

You get the most important structural change in innovation finance since the invention of the limited partnership.

When a network of thousands of startups, investors, customers, and domain experts can share signals through AI-mediated coordination, the information advantage that justified Sand Hill Road’s 2-and-20 fee structure dissolves. The network becomes the fund. The fund becomes the network.

This is not speculation. The architecture already exists. Y-Combinator proved that a standardized mentorship network could produce better outcomes than traditional VC at the early stage. AngelList proved that syndicates could aggregate small checks into competitive rounds. Rise of the Rest has demonstrated that emerging VCs outside traditional hubs can identify deals that coastal firms miss entirely.

These are pieces of a system that has not yet been assembled. AI is the assembler.

The Productivity Unlock

Forget the conservative forecasts. The economists projecting 0.2 percentage point TFP gains are measuring the wrong thing. They are measuring AI as a tool applied to existing industry structures. They should be measuring AI as a solvent that dissolves the organizational barriers between customer needs and innovative solutions.

The U.S. ran 4% GDP growth in the 1940s through the 1960s. Not because of government policy. Because Vannevar Bush’s wartime innovation infrastructure spilled into the hands of entrepreneurs in every state, creating new businesses and expanding the economy in every direction. Distributed innovation, fueled by distributed capital, accessing distributed knowledge.

Then the system concentrated. Banks consolidated lending. VC concentrated in a few zip codes. Innovation anchored to specific geographies. The coordination that was so powerful in the middle of the twentieth century weakened.

Network-centric innovation reverses that concentration. When AI agents can negotiate with suppliers, AI tools can prototype products, and AI-mediated networks can match founders with domain-specific investors and customers regardless of geography, you get something that looks a lot more like 1955 than 2015.

Except faster. Because the constraint on post-war innovation was the speed of human communication. Letters, phone calls, conferences, travel. The constraint on network-centric innovation is the speed of digital communication. Measured in milliseconds.

We spend $1.9 trillion per year in the U.S. on healthcare costs related to poor nutrition. That number is not a policy problem. It is an innovation problem. It persists because the entrepreneurs who could solve it cannot access the capital, customer signals, and distribution networks they need. The VC model sends that capital to AI infrastructure companies in San Francisco, not to food-as-health startups in the Midwest where the customers and agricultural talent actually live.

Network-centric innovation does not require choosing between AI infrastructure and applied innovation. It funds both, because the network exposes the information arbitrage in both. And when AI reduces the cost of building a company by 80%, the capital required to fund a food-as-health startup drops from $10 million to $2 million, and the addressable market for innovation capital expands by an order of magnitude.

What to Watch

The shift from hierarchical to network-centric innovation will not arrive as a single event. It will arrive as a series of anomalies that the existing model cannot explain.

Watch for profitable startups that never raise a traditional Series A. Watch for venture returns from non-traditional geographies that exceed coastal benchmarks. Watch for AI-native companies that reach $10 million in revenue with fewer than 10 employees. Watch for investor networks that outperform traditional VC funds by aggregating distributed signals rather than relying on partner intuition.

Watch for total factor productivity gains in sectors that have been flat for decades. Agriculture. Construction. Education. Healthcare. These are the sectors where most Americans work and most GDP is produced. They have been ignored by the innovation economy for a generation. They will not be ignored much longer.

The incumbents in each of these sectors have used their lobbying leverage to protect their industries from competition. Stock buybacks signal that management has no better use for capital than to inflate their own options. When AI reduces the cost of competition by orders of magnitude, that protection dissolves. Not through policy. Through economics.

The venture capital industry as currently structured is a $700 billion relic of Sloan-era organizational logic applied to a network-centric world. The entrepreneurs and investors who see the structural change will build the companies that define the next 30 years. The ones who do not will keep writing checks to AI infrastructure companies and wondering why their returns look like the S&P 500.

Innovation is still a deflationary force. It still improves society’s productivity and resource allocation more efficiently than wars or politics. The process of a few entrepreneurs succeeding and many failing is still the fastest path to a higher standard of living.

What changes is the structure through which that innovation flows. The hierarchy gave way to the network. The fund gives way to the network. Capital, talent, customers, and information find each other at the speed of light instead of the speed of a partner meeting.

The $4 trillion venture economy is not a prediction. It is a description of what happens when you remove the friction between money and opportunity at scale. AI does not create that future. AI reveals it. The knowledge was always there, scattered across a continent of entrepreneurs, customers, and investors who could never coordinate fast enough to act on it.

Now they can.