AI accounting startup Rillet just became a unicorn in 48 hours—a fundraising sprint that would have seemed impossible two years ago when foundation model companies dominated venture attention. CEO Nicolas Kopp presented growth metrics at a board meeting, and the response was immediate: Iconiq, Sequoia, and other top-tier firms rushed to commit capital without the extended due diligence that typically precedes nine-figure rounds. The speed signals a fundamental reallocation of venture dollars. This year, 250 companies have joined the unicorn ranks through mid-August, up 30% from 2025's 193 companies. While that growth looks broad, the sectoral breakdown reveals the real story: robotics, AI infrastructure, healthcare applications, financial services tools, and AI deployment platforms are dominating, not general-purpose foundation models.
The contrast with 2024-2025's funding landscape is stark. Foundation model companies—those building large language models or core AI systems—typically took 6-12 months to raise at scale, navigated intense technical scrutiny, and faced capital concentration among a handful of mega-rounds. Rillet's two-day close happened because investors see a different risk profile in vertical AI: proven revenue models, defensible customer relationships, and immediate enterprise adoption. The startup isn't competing against OpenAI or Anthropic; it's automating a specific workflow that accountants have paid for traditionally. This is venture math that works. Other wins this week underscore the pattern: AI inference startups landed substantial funding, alongside narrower plays in video creation, voice-to-text, and even AI for construction bidding and recycling optimization. Each targets a specific pain point rather than attempting to build foundational technology.
For investors, the vertical AI thesis offers clarity. A venture partner backing Rillet can point to gross margins, customer retention metrics, and a clear path to profitability—the same signals that matter in traditional software. Foundation models, by contrast, remain capital-intensive research bets with uncertain unit economics. As unicorn velocity accelerates and capital pours into domain-specific applications, the market is essentially declaring that the era of funding pure AI research through venture rounds is waning. The money has moved downstream, to startups that combine domain expertise with AI capabilities. For founders and LPs, the signal is clear: specificity beats generality in 2026's funding environment.