Oct 8, 2025, 8:17 AM ETConference Presentation, Fireside Chat, Panel

Fractile AI, Rogo, Scribble VC, San Francisco Compute: The AI Trillion-Dollar Opportunity

Eric Manisse (CEO, San Francisco Compute): Operates a marketplace and software platform to democratize compute accessibility globally.
Elizabeth Weill (Founder/Managing Partner, Scribble Ventures): Runs an early-stage, AI-focused fund with an $80 million vehicle, specializing in pre-seed and seed investments in top-tier founders.
Walter Goodwin (Founder/CEO, Fractile): Building novel silicon and a platform designed to run frontier AI models 50 times faster and 10 times cheaper than current flagship GPUs.
John Willett (Co-founder/COO, Rogo): Developed an AI platform fine-tuned for large financial services firms (e.g., Lazard, JPMorgan) to automate high-frequency analyst tasks.

Adopt a "Ship Fast, Fail Fast" software mindset: San Francisco Compute rebuilt its marketplace three to four times to secure immediate customer traction, prioritizing speed over initial structural perfection.
Bottom-up adoption for enterprise sales: Rogo initiated growth by targeting junior analysts to create internal virality before targeting high-level decision-makers for pricing and security approvals.
Calibrated speed for hardware startups: Fractile accepts a 2-year timeline from architecture definition to tape-out, prioritizing market fit over speed to avoid a $50 million loss on unwanted silicon.
VC advice on reinvention: Scribble Ventures urges founders to assume they will be using the "dumbest models they will ever use" and to rebuild products daily to stay ahead of rapid model evolution.
The "Crystall Ball" challenge: Hardware founders must forecast trends 2–3 years into the future to align product definition with customer needs by the time silicon ships.

North Star Metrics beyond revenue: Fractile identified early traction by tracking customers proactively seeking debugging help, validating demand before scaling.
Hybrid deployment for security compliance: Rogo balances InfoSec constraints (keeping apps within the client's perimeter) with SaaS R&D needs (hosting model inference) via hybrid architecture.
Hallucination risk mitigation: Rogo enforces strict user training, teaching bankers to treat the AI as an "intern" and audit every citation to prevent trust loss in financial advice.
Financial governance for marketplaces: San Francisco Compute prioritizes securing custody of money and time as the primary internal governance control.
Forward-looking regulatory influence: Both San Francisco Compute and Fractile actively collaborate with California state bodies and US national export control teams to shape regulations rather than merely comply.

Small teams do not scale indefinitely for deep tech: Walter Goodwin and Eric Manisse argue that while AI accelerates non-technical functions, deep hardware and infrastructure roles still require specialized, human expertise.
Specialized sub-teams remain essential: Rogo maintains distinct sub-teams (product, research, infra) because AI tools are less effective for novel research compared to solving existing engineering problems.
Non-technical efficiency gains: AI tools allow startups to delay hiring specialists in legal, data analytics, HR, and customer support, enabling leaner operations in the early years.
Orchestration complexity management: AI aids in synchronizing large, fast-growing organizations by centralizing documentation and governance, mitigating the quadratic complexity of human management.

Downside risk modeling is critical: Founders are advised to over-model downside risks and anticipate market volatility (e.g., GPU price fluctuations) to ensure longevity over rapid, fragile growth.
Data and domain specificity as moats: Rogo's strategy relies on comprehensive, heterogeneous data sources and domain-specific fine-tuning to create value that persists beyond generic model capabilities.
Velocity vs. durability: Successful companies must balance high-velocity execution with the development of "hard" barriers, such as patents and deep technical complexity, to prevent overnight commoditization.
Founder centrality: Scribble Ventures views the top 1% of founders as the primary differentiator for long-term survival in an environment where tools and models change rapidly.

Export controls as an opportunity: Fractile views the separation of global supply chains (US, Europe, China) as a chance to build domestic stacks for energy, data centers, and silicon.
Data residency constraints: Rogo must navigate strict EU and US data residency laws, particularly regarding non-public information (MNPI) in financial contexts.
Pro-regulatory stance: Startups argue that engaging with regulators early creates a clearer operating environment, contrasting with the "skirt around controls" approach seen in the crypto era.
Government adoption signals: Increased adoption of AI by government agencies (DC) indicates a collaborative "build" mindset rather than a fear-based regulatory crackdown.

Transformative sectors identified: Elizabeth Weill predicts breakthroughs in personalized education (AI tutors), health (personalized prescriptions), robotics, and space exploration.
Shift in monetization units: Fractile anticipates a transition from pricing based on "tokens" to pricing based on "quality of service" as the primary unit of compute consumption.
Human-centric value in finance: John Willett argues AI will commoditize low-level analysis, increasing the value of senior bankers' human skills (networking, market intuition) similar to how Excel shifted but did not eliminate investment banking roles.
Compute as the new oil: San Francisco Compute views the next industrial revolution as the transition of compute from a scarce resource to an on-demand utility via a global "compute grid."
Historical parallels: The panelists draw parallels to the Industrial Revolution and the introduction of electricity, suggesting AI will fundamentally reshape global labor dynamics and infrastructure.