The current market is defined by a stark valuation dichotomy, with undervalued public SaaS companies and overvalued private AI firms [1, 17].
Venture capital LPs are facing a critical liquidity shortage, or 'DPI problem,' which needs to be resolved through increased M&A and IPO activity [3, 15].
The AI application layer is undergoing rapid, disruptive turnover, and will ultimately consolidate around a few dominant players in each category [18, 24, 29].
The foundational model lab category is generally overfunded and overvalued, suggesting better investment opportunities exist elsewhere in the AI stack [23].
A long-term, thesis-driven approach, like Madrona's 2016 shift to 'intelligent applications,' is crucial for navigating the hype and volatility of the current AI market [20].
2013
McIlwain notes the origin of the Lean programming language at Microsoft Research, a technology later central to the thesis of portfolio company Axiom [12].
2016
Madrona Venture Group, where McIlwain is a key figure, strategically shifted its investment thesis to focus exclusively on 'intelligent applications' infused with data and machine learning [20].
Summer 2022
McIlwain reveals that Madrona Venture Group made the significant decision to pass on investing in OpenAI's round at a $25 billion valuation [28].
~3 years prior to 2026 outlook
McIlwain provides a baseline for portfolio company Echodyne's growth, noting it had $20 million in bookings [27].
Last year (relative to 2026 outlook)
He highlights Echodyne's massive growth, with bookings soaring to $340 million [27]. He also notes the high turnover on the Intelligent Application 40 list, with 27 new companies appearing [18, 29].
Present (2026 Outlook)
McIlwain characterizes the current market as having a 'massive DPI problem' for LPs [15], a stark valuation dichotomy between public and private tech [17], and an overfunded foundational model category [23].
▶The Great Valuation DivideApr 2026
McIlwain frequently contrasts the depressed valuations of public software companies with the sky-high valuations of private AI firms. He points to examples like ServiceNow trading at record lows while AI unicorns raise massive rounds, creating a challenging environment for investors to navigate.
This valuation gap suggests a market dislocation where future growth expectations for AI are being priced in aggressively, creating both significant opportunities and risks for venture capitalists who must decide where to place their bets.
▶The AI Application Layer ShakeoutApr 2026
McIlwain emphasizes the rapid disruption and consolidation occurring in the AI application space. He cites the high turnover on the Intelligent Application 40 list as evidence of this volatility and predicts that most categories will consolidate to just two or three winners.
For investors, this signals that backing a company in a crowded AI application category is a high-risk, high-reward proposition, where market leadership is fragile and long-term dominance is far from guaranteed.
▶Venture Capital's Liquidity CrisisApr 2026
A core concern for McIlwain is the 'massive DPI problem' where Limited Partners (LPs) are not seeing cash returns from their venture investments. He connects this to a slow M&A and IPO market, which has trapped capital in late-stage private companies.
This focus on DPI highlights a shift in the venture landscape from prioritizing paper markups to demanding real, distributed returns, which will likely increase pressure on VCs to find exit opportunities for their portfolios.
▶Madrona's 'Intelligent Application' ThesisApr 2026
McIlwain details his firm's strategic pivot in 2016 to focus exclusively on 'intelligent applications' infused with data and machine learning. This thesis is exemplified by successful portfolio companies like Echodyne and the sale of Statsig to OpenAI, demonstrating a long-term commitment to the AI-native ecosystem.
Madrona's early and disciplined focus on a specific AI thesis appears to be a key differentiator, allowing them to build deep expertise and potentially outperform in a crowded market, even if it means passing on seemingly obvious deals like OpenAI.