Is Anthropic Making the Biggest Mistake in AI History | E2258

What It Covers

Jason Calacanis and guests Logan Allen (Fin Capital) and Eric Voorhees (Venice AI) cover Anthropic's revenue surge to $19B annualized run rate, the Pentagon contract dispute, OpenClaw's rise as the most-starred GitHub project in history, stablecoin adoption signals, AI-driven venture capital frameworks, and two startup demos focused on agentic web analytics and OpenClaw accessibility tools.

Key Questions Answered

• •Anthropic Revenue Trajectory: Anthropic's annualized revenue run rate reached $19B in early 2026, up from $9B at end of 2025 and $14B just weeks prior. Claude Code alone accounts for approximately $2.5B of that figure. Enterprise relationships represent over 80% of Anthropic's revenue, compared to under 5% for OpenAI, making enterprise-first orientation a measurable competitive differentiator in AI monetization strategy.
• •AI Adoption Signal Framework: A three-stage adoption signal framework helps identify breakout technologies: first, tinkerers and hackers adopt it; second, startups integrate it; third, the evaluator personally begins using it obsessively. All three signals have now converged around OpenClaw and agentic AI tools, mirroring the pattern seen with Venmo and PayPal when fee-sensitive users like contractors began preferring them over card networks.
• •Venture Firm AI Transformation: The future venture firm structure mirrors the old Benchmark model — a small number of senior partners with deep relationships, supported by hundreds of AI agents handling deal sourcing, portfolio monitoring, founder research, and theme analysis. Fin Capital built a founder DNA scoring model using 40 weighted criteria, with repeat founder status, CEO tenure duration, and exit track record (even small exits) as the highest-predictive variables.
• •Stablecoin Adoption Indicator: Stablecoin mainstream adoption follows a predictable sequence: first criminal use, then discreet but legal activity like offshore wagering, then fee-sensitive small business operators such as contractors and gardeners seeking to avoid the 3–5% card network fees. When that third group begins requesting stablecoin payments, the technology has crossed into mainstream viability — a threshold Circle and USDC are now approaching with the Clarity Act providing regulatory tailwinds.
• •Privacy-First AI as Product Differentiation: Venice AI offers private, uncensored inference with no conversation logging, positioning against ChatGPT and Claude on user data sovereignty. Its token model (VVV) generates sub-tokens called DMs, each providing $1 per day of inference credit as a perpetuity-style asset. For users handling legal contracts or medical records, running documents through a private inference layer eliminates the risk of sensitive data entering centralized training pipelines permanently.
• •Agentic Web Preparedness Gap: Most websites are built for human UI and fail when AI agents attempt to crawl or interact with them, creating a measurable gap in agentic traffic attribution. Sightline tracks agent visits server-side, where traditional analytics tools miss them entirely. Businesses should audit whether their sites are agent-readable, consider Cloudflare's pay-per-crawl model for IP monetization, and build attribution funnels that connect agent research visits to downstream human conversion events.