Venture capital has always been a game of duration. AI is accelerating innovation, but it may be compressing the very monopolies that historically powered the industry’s largest returns.

Why AI Will Compress Venture Returns Before It Expands Them

For the better part of two decades, venture capital has operated on a relatively simple assumption: technological breakthroughs create temporary monopolies, and temporary monopolies create outsized returns. The formula was never foolproof, but it worked often enough to build venture dynasties and mint a small army of billionaires.

Beneath that formula sat an even more important assumption: competitive advantages compound faster than competitors can erase them. Venture capital depends on time behaving predictably. Capital enters a market, leaders emerge, moats deepen, and ownership stakes mature into extraordinary outcomes.

AI may be the first major technology wave in decades to challenge that sequence. For the first time in a generation, the cost of imitation is falling almost as quickly as the cost of creation. The same tools allowing startups to build faster are allowing competitors to catch up faster. When competitive advantages decay faster than ownership stakes mature, the power-law structure underpinning venture returns begins to wobble.

Yet the dominant narrative around AI remains one of acceleration. Startups are scaling faster. Products are shipping faster. Revenue milestones are arriving earlier than they did during the SaaS era. A pair of developers armed with foundation models, cloud infrastructure and a sufficient supply of caffeine and salty snacks can now build capabilities in a matter of days that once required entire floors of engineers and years of development.

Investors see that velocity and instinctively assume it translates into larger outcomes. That assumption deserves more scrutiny than it is receiving. This is because velocity cuts both ways. The same force accelerating creation is also accelerating imitation. AI is compressing the time between innovation and commoditization.

Competitive advantages that once endured for years are increasingly measured in quarters. The result may be an ecosystem that creates meaningful companies at unprecedented speed while simultaneously shortening the duration of their economic dominance.

And duration, more than most people in venture capital care to admit, has always been the real engine behind extraordinary returns.

The Hidden Variable Behind Venture Outcomes

The mythology of VC romanticizes invention. Founders are cast as lone visionaries standing at the edge of the technological frontier, dragging civilization forward through intellect, conviction, and sheer force of will. However, reality is both far less cinematic and more relevant to investment returns.

Most legendary venture outcomes were built not on invention alone, but on sustained asymmetry. Google did not become a trillion-dollar company simply because its search algorithm was superior in 1999. Google was famously the seventeenth search engine to enter the market. Google became Google because it maintained structural dominance in search for more than two decades.

Microsoft followed a similar path. Windows was hardly beloved software in its early years. It was unstable, inelegant, and frequently mocked by developers. Yet Microsoft built an ecosystem so deeply embedded that competitors struggled to dislodge it for a generation.

Technology history is full of companies that were ‘first movers’ but still lost. Netscape innovated faster than Microsoft. Palm built a devoted following years before the iPhone debuted in 2007. MySpace achieved escape velocity before Facebook became culturally dominant. Indeed, the highways and byways of Silicon Valley are littered with the carcasses of companies that reached the summit first only to discover the summit was sitting on a fault line.

The venture industry consistently over-credits innovation and under-credits durability. Markets rarely reward whoever arrives first. They reward whoever becomes hardest to replace once distribution, customer behavior, data, and infrastructure begin reinforcing one another.

That dynamic defined much of the SaaS era. Salesforce, Adobe, ServiceNow, and Atlassian benefited from long periods during which switching costs accumulated faster than competitors could dismantle them. Their success was not merely a function of building better software. It was a function of maintaining advantage long enough for those advantages to compound.

In previous technology cycles, building a credible competitor required years of engineering effort, substantial infrastructure investment, and large teams. Today, the cost and speed of replication have collapsed. A startup demonstrates traction in an AI category and competitors appear almost immediately, each offering marginally different workflows, interfaces, or orchestration layers. Entire categories increasingly resemble a knife fight in a closet conducted at machine speed.

The venture ecosystem still behaves as though breakout growth automatically buys years of competitive breathing room. Increasingly, it does not. The distance between market leader and market follower is shrinking, and the period between innovation and imitation is compressing right alongside it.

AI’s Brutal Compression Cycle

One of the defining characteristics of the current AI cycle is how quickly differentiation spreads through the market.

Image generation offers an instructive example. In 2022, companies like Midjourney, Stability AI, and OpenAI appeared to hold meaningful advantages in a category many expected would consolidate around a handful of dominant players. Instead, open-source alternatives proliferated, quality rapidly converged, and pricing pressure intensified. The technology improved at breathtaking speed even as the economic defensibility of individual participants weakened.

That pattern is increasingly visible across the broader AI landscape. A startup introduces breakthrough functionality, capital floods into the category, competitors rapidly replicate the core experience, and differentiation migrates away from the model itself toward distribution, proprietary data, workflow integration, or enterprise relationships. The technology keeps improving. Defensibility does not.

The coding-assistant market already offers a glimpse of how quickly competitive advantages can compress in the AI era. GitHub Copilot established an early lead and appeared well positioned to dominate the category. Yet startups such as Cursor rapidly accumulated developer loyalty and forced incumbents into an accelerated cycle of product iteration. The market continues to grow at an extraordinary pace. Yet the real question is no longer who can build the best product. It is who can remain meaningfully different once competitors can replicate core functionality in months rather than years

This is where much of the current AI discourse becomes oddly superficial. The market remains obsessed with capability expansion while paying far less attention to durability compression.

Recent research has documented dramatic declines in inference costs, token pricing, and the cost of achieving benchmark performance. Those developments are enormously positive for customers and for society at large, but they are considerably less comforting for investors underwriting long-term pricing power.

As intelligence becomes cheaper and more abundant, many AI products begin to resemble utilities more than monopolies. Buyers become increasingly willing to switch providers. Enterprise software has historically benefited from organizational inertia. Companies tolerated mediocre products, rising prices, and sluggish innovation because replacing critical systems was painful, risky, and expensive. AI may weaken some of that inertia by reducing implementation friction and accelerating feature convergence. If software becomes easier to replace, the durability assumptions underpinning many SaaS valuations begin to weaken alongside it.

Public markets have already begun repricing software businesses whose durability assumptions are increasingly under pressure. Margins compress. Switching costs erode. Competitive advantages migrate away from the model layer and toward infrastructure, data, distribution, and workflow ownership. In many cases, the AI itself ceases to be the moat.

That reality also helps explain why incumbents suddenly look far more formidable than many AI-native founders expected. When intelligence becomes abundant, distribution regains importance with remarkable force. Microsoft can place AI inside Office. Salesforce can weave it directly into enterprise workflows. Google can expose generative search to billions of users almost overnight.

This serves as more evidence that while startups may move faster, the incumbents still own the roads.

The Open-Source Problem Venture Capital Keeps Underestimating

The venture industry has always favored businesses that become stronger as their ecosystems mature. Open-source AI complicates that dynamic in ways many investors are still underestimating.

Bill Gurley has spoken extensively about the strategic implications of open-source AI and the pressure it places on traditional software economics. China’s increasingly aggressive open-model ecosystem only accelerates the trend. Competitive advantages that once depended on proprietary access to cutting-edge models now face constant erosion from globally distributed research communities moving at extraordinary speed.

That creates an uncomfortable reality for many venture-backed AI startups. A significant portion of today’s application layer consists of companies building on top of foundation models that are becoming more capable, cheaper, and more widely available with each passing quarter. Many will build real businesses but far fewer will build durable monopolies.

That distinction matters because venture returns have never been driven by popularity alone. They are driven by sustained asymmetry. A company reaching $50 million ARR in record time is impressive. A company that can defend its economics for fifteen years is what creates legendary venture funds.

Periods of technological euphoria often obscure that difference. Venture capital has a long history of confusing speed with inevitability.

The dot-com era offers a useful reminder. The internet transformed the global economy beyond recognition, yet hundreds of venture-backed companies disappeared as distribution advantages collapsed and infrastructure became commoditized. AI may compress that entire cycle into a fraction of the time.

At its core, venture capital has always depended on scarcity somewhere in the system: scarce capital, scarce talent, scarce infrastructure, scarce distribution, or scarce technology. Open-source AI attacks several of those constraints simultaneously. Yet much of the market continues to underwrite AI startups as though proprietary technology alone guarantees durable pricing power.

Faster Growth May Produce Smaller Outcomes

One of the more uncomfortable possibilities in AI is that the industry may produce more billion-dollar companies while simultaneously producing fewer enduring monopolies.

That sounds contradictory until you consider what AI is actually doing to the economics of company formation. Barriers to building software continue to fall. Products reach market faster. Teams remain smaller for longer. Capital efficiency improves. As a result, more companies are likely to achieve meaningful scale than in previous technology cycles. Early evidence already points in that direction, with AI startups attracting extraordinary amounts of capital while producing some of the fastest revenue ramps the industry has ever seen. The problem is that speed does not merely accelerate growth. It accelerates competition as well.

Categories become crowded faster, product differentiation narrows faster, and customer expectations evolve faster. Entire markets can move from breakthrough innovation to margin pressure before many participants have fully matured operationally. What appears to be a durable lead can increasingly turn out to be a temporary head start.

Scott Galloway recently warned that current AI valuations may be embedding overly optimistic assumptions about long-term defensibility. His broader point deserves attention because venture capital ultimately monetizes sustained market power, not technological novelty.

As discussed earlier in this piece, a company on a torrid growth trajectory is impressive but one that can maintain pricing power and market position for a decade or more is the whole ballgame for venture investors. Those are not necessarily the same businesses.

This distinction sits at the heart of venture economics. Investors have historically tolerated extraordinary upfront losses because dominant software companies often enjoyed a decade or more of expanding margins after category consolidation. The payoff period was long enough to justify the risk.

However, if AI shortens the period between breakout growth and margin competition from ten years to three, terminal outcomes begin to look very different. A company that reaches $300 million ARR in record time may still become a successful business. Yet if pricing pressure, feature convergence, and competitive saturation arrive years earlier than expected, the outcome investors ultimately underwrote may never materialize.

Where Durable Value May Actually Persist

If the AI era ultimately produces durable winners, they are likely to emerge from the ownership of bottlenecks rather than the ownership of intelligence itself.

Generalized intelligence is becoming more abundant every quarter. Bottlenecks remain scarce. Compute capacity, energy availability, inference optimization, enterprise distribution, regulatory trust, proprietary data, and deeply embedded workflows all possess characteristics that are considerably harder to commoditize than the models running on top of them.

This helps explain why infrastructure remains so strategically important. Nvidia’s continued dominance illustrates how value often concentrates around foundational layers during periods of technological transition. The companies controlling critical inputs frequently capture more durable economics than those competing to build the latest application on top of them.

The same principle extends beyond infrastructure. Companies deeply embedded within enterprise workflows may retain meaningful advantages because trust, compliance, operational integration, and switching costs create friction that technology alone cannot easily eliminate. Likewise, firms controlling proprietary real-world datasets or highly specialized operating environments may prove more defensible than application-layer businesses competing primarily on interface design and feature velocity.

Secondary markets may be the first place that repricing becomes visible. As competitive windows shorten, investors become less willing to underwrite terminal values that sit a decade in the future. Growth equity investors and crossover funds may increasingly evaluate AI companies less like compounding software franchises and more like high-velocity assets whose competitive position must be constantly re-earned.

None of this means AI will fail to produce extraordinary companies. Quite the opposite. The next decade will almost certainly create enormous winners. But fewer of those winners may enjoy the luxury that defined many of the great software companies of the last generation: years of relatively uncontested dominance during which advantages could compound faster than competitors could erase them.

Final Word

The venture industry has largely framed AI as an acceleration story. And it is. Products are built faster, companies scale faster, and markets emerge faster than anything we witnessed during the SaaS era. But AI is also a compression story.

The same forces accelerating innovation are compressing the lifespan of competitive advantage. Product differentiation diffuses faster. Competition arrives sooner. Categories saturate earlier. What once looked like a durable moat increasingly resembles a temporary lead that can erode quickly.

That distinction matters because venture returns have never been determined solely by how quickly a company grows. They have been determined by how long that company can sustain advantages that competitors struggle to replicate. The software era produced extraordinary outcomes because many market leaders enjoyed years, and sometimes decades, of relatively uncontested dominance. AI appears poised to shorten those windows.

For founders, that means building a great product may become less important than building durable distribution, proprietary data, workflow ownership, or some other source of defensibility that survives once the underlying technology becomes widely available.

For investors, it means paying closer attention to the duration of advantage than the speed of growth.

Silicon Valley spent the last twenty years celebrating software eating the world. The next generation of software companies may spend less time defending moats and more time defending attention.

***Jonathan Tower has been a global venture investor for over 20 years, having managed more than $5 Billion in AUM, invested in more than 85 companies, and seeded 9 companies that went on to become unicorns across three core investment themes: consumer (marketplaces, ecommerce enablement, digitally native brands), enterprise (software, services, infrastructure, storage, data orchestration) and frontier technologies* (AI/ML, IoT, robotics, Fintech, etc).

Jonathan’s direct investments have resulted in more than $10 Billion in exits, including early bets in Jet.com (acquired by Walmart for $3.5 Billion), Dollar Shave Club (acquired by Unilever for $1 Billion), Freshly (acquired by Nestle for $1.5 Billion), IfOnly (acquired by Mastercard), InsideView (acquired by Demandbase), and MapR Technologies (acquired by HP). Other notable investments, which Jonathan led or helped champion, include Groq (acquired by Nvidia for $20 Billion), Hammerspace, Cohere, TogetherAI, Snorkel AI, Jeeves, SingleStore, Artera, Cart.com, Madison Reed, Qumulo, and many other companies that have gone on to become market leaders.

Jonathan writes frequently on venture capital and technology topics on his blog, Adventure Capitalist, and he’s been a frequent contributor to The New York Times, Fortune, The Wall Street Journal, FastCompany, Forbes, The Washington Post, The LA Times, and other leading publications.

X: @jonathan_tower; instagram: jonathan_tower