Venture capital today is no longer a uniform playing field. Instead, it resembles a barbell—heavy at the ends and light in the middle. On one side, early-stage startups, particularly those native to Generative AI, are gaining momentum and capital with relative ease. On the other, late-stage companies that once thrived in an era of easy money are now struggling to scale efficiently under compressed valuations and tighter capital flows.
This bifurcation isn’t just about stage or sector—it reflects deeper shifts in how innovation is funded, how markets respond to disruption, and how value is created in an AI-first world. Drawing on decades of experience and observations across multiple venture cycles, Matt McIlwain of Madrona Venture Group offers a compelling framework for understanding today’s venture environment—and where it’s headed next.
The Barbell Effect in Venture Capital
McIlwain describes the current VC climate as a “tale of two cities”—or more aptly, a “barbell.” On one end are late-stage startups that raised funds during the heady days of 2019–2021. Many of these companies are now struggling to scale in a capital-efficient way. They face compressed valuations, uncertain exits, and reduced investor enthusiasm.
On the other end of the barbell are early-stage startups, especially those built around Gen AI from the ground up. These “AI-native” firms are garnering significant attention and capital due to their alignment with emergent technologies and cost-effective scalability. The difference is stark: late-stage companies are in triage, while early-stage Gen AI startups are in acceleration.
Disconnect Between Public and Private Markets
A key challenge today is the widening disconnect between public and private market valuations. Public markets reprice daily and saw software multiples soar to 15x in 2021 before correcting to around 5x. Private markets, by contrast, adjust only when a new funding round occurs—many of which haven’t happened since 2021.
This valuation lag creates misalignment, confusion, and occasionally, overpricing. McIlwain cautions investors to avoid mistaking “book value” for market reality, particularly in the realm of late-stage investing.
The Strategy Behind AI Partnerships
Big Tech’s strategic alliances with AI model developers—Microsoft with OpenAI, Amazon with Anthropic, Oracle with Cohere—are not merely about access to compute resources. These partnerships serve a dual purpose:
Cloud Infrastructure Leverage: By offering AI models via services like Amazon Bedrock or Azure, cloud giants ensure their platforms remain central to AI development.
Market Shaping: Through co-investment and go-to-market support, these firms influence not just the technology stack, but also the distribution landscape for Gen AI applications.
What’s emerging is a marketplace approach, where multiple models—from OpenAI to Meta’s LLaMA to Stability.AI—coexist on cloud platforms, empowering developers with choice and flexibility.
Seattle’s Edge—and Its Gaps
Seattle, McIlwain argues, remains a wellspring of technical talent, especially in AI. From the early days of AWS and Azure to Madrona’s investment in Turi (later acquired by Apple), the city has been ahead of the curve. However, it remains underfunded at the local level. Most co-investors in promising startups still come from outside the region.
To bolster the ecosystem, there’s a need for more local early-stage capital and smoother transitions for professionals moving from large enterprises like Microsoft or Amazon into startup environments.
Intelligent Applications and the “Mini Model World”
One of the more compelling insights from Madrona’s Intelligent Applications Summit is the idea of a “mini model world.” Rather than relying on a monolithic AI model, startups are building composite “model cocktails” tailored to specific use cases. This ensemble approach enables better performance, customization, and faster time-to-value.
Still, the industry is largely in a prototype phase. Full production deployments of Gen AI remain rare, meaning we’re still early in this innovation cycle.
Advice for Aspiring VCs and Founders
McIlwain leaves us with grounded advice: align with people who inspire and challenge you, pursue what you’re genuinely passionate about, and always assess “founder-market fit” before diving into a startup idea.
For new VCs or founders, understanding the probabilistic nature of startup success is essential. Drawing from Annie Duke’s Thinking in Bets, he emphasizes making informed decisions without perfect data—a theme that resonates deeply in today’s fluid market.
The rules of venture investing are being rewritten, but the fundamentals—disciplined capital allocation, deep technical understanding, and long-term thinking—remain more vital than ever.
Nataraj is a Senior Product Manager at Microsoft Azure and the Author at Startup Project, featuring insights about building the next generation of enterprise technology products & businesses.
Listen to the latest insights from leaders building the next generation products on Spotify, Apple, Substack and YouTube.