Conference Presentation, Panel
2035 Decoded: Navigating the Decade Ahead | Dust, Glean & More | RAISE Summit 2026
- Time Horizon & Context: Enterprise AI adoption is projected to be a 10-to-15-year journey, requiring strategic decisions made today with a 2035 outlook rather than short-term predictions.
- Market Capitalization Trends: Public hyperscalers are aggressively increasing AI CapEx; Meta is estimated to raise its 2027 CapEx investment from $75 billion (2025 levels) to between $250 billion and $300 billion.
- Architectural Reality: Current AI success requires a multi-layered ecosystem combining foundational models, "harnesses" (prompting frameworks), and systems for generating new knowledge to create a feedback loop.
- Model Specialization: Large Language Models (LLMs) are effective for language and creativity but lack formal reasoning capabilities; "Energy-Based Models" (EBMs) are emerging specifically for spatial reasoning, prediction, planning, and mission-critical sectors like finance and robotics.
- Enterprise Strategy: Leaders are advised to avoid committing to specific models or fully decentralized architectures, instead prioritizing "context graphs" and a reference architecture that allows for vendor switching.
- Intellectual Property: Enterprises must retain ownership of compounding AI learnings and memory infrastructure to maintain competitive advantage, rather than ceding this data to external tech vendors.
- Talent Density: Success in the agentic era depends on hiring talent with high "systems thinking" and adaptability; the leverage of a single engineer with access to agents is orders of magnitude higher than in the past.
- Risk & Safety: The shift from deterministic to non-deterministic tools introduces hallucination risks, necessitating formal verification and constraint mechanisms for high-stakes applications like banking and autonomous driving.
- Value Dislocation: Panelists predict a massive divergence in value among the ~391 non-semiconductor S&P 500 companies, with approximately 125 emerging as winners and 250 losing significant value due to the inability to transition to agentic workflows.
- The Moat: "Context" (deep, defensible data) is identified as the primary competitive moat, allowing companies like FedEx, JP Morgan, and AXA to translate existing organizational knowledge into agentic superpowers.
- Organizational Structure: Organizations will likely remain hierarchical but flatter, with wider spans of control, as humans design workflows and use AI agents as tools rather than replacing human hierarchy entirely.
- Job Evolution: Repetitive white-collar tasks (e.g., coding, underwriting) face significant automation, but historical precedents suggest this will drive productivity gains and hiring in high-leverage areas rather than net job elimination.
- Economic Impact: The transition may hollow out the "modest middle" of the market, concentrating value at the top where data/talent density compounds and at the bottom where friction for serving local niches is removed.
- Final Valuation Metric: Long-term value will accrue to entities that own the system specifications and constraints, keeping human oversight central to defining AI behavior.