AI in 2026: 3 Predictions For What’s To Come (a16z Big Ideas)

Oliver Hsu (AI & Autonomous Science)

• Core Thesis: Advances in AI reasoning and robot learning are transitioning scientific discovery from human-led to collaborative or fully autonomous systems.
• Evolution of Lab Automation:
  • Traditional lab automation is limited to pre-programmed physical motions; the new frontier combines this with AI-driven experiment planning and reasoning.
  • Near-term reality involves human-AI collaboration where systems record interpretability data to explain experimental iterations.
  • "Fully self-driving science" (a closed loop of self-iterating AI and experimentation) is defined as a long-term destination rather than current capability.
• Technical Prerequisites:
  • Progress requires maturation in mathematical reasoning, physical reasoning, simulation, world models, and robot learning.
  • Capabilities currently exist but are unevenly developed, requiring synchronization before full closure of the autonomous loop.
• Market Adoption Dynamics:
  • Adoption will be driven by sectors with mature demand for research outputs, specifically life sciences/pharma, chemicals, and material science.
  • Speed, capability, and cost advantages are most valuable where "ready and willing buyers" for research outputs exist.
• Ecosystem Landscape:
  • Startups: Early-stage activity includes Medra (life sciences), Chemify and Yoneda Labs (chemistry), and Periodic Labs (autonomous science).
  • Public-Private Collaboration:
    • The Department of Energy's "Genesis" mission unites academia, national labs, and AI firms for AI-driven science.
    • DeepMind announced a partnership with the UK government for scientific discovery collaboration.

Brian Kim (AI Consumer Connectivity)

• Strategic Shift (2026): Major consumer AI products are pivoting from "productivity" (working) to "connectivity" (staying connected and being seen).
• User Behavior Trends:
  • AI is shifting mindshare from traditional tools to platforms that facilitate digital relationships and validate personal identity.
  • Users are increasingly willing to share their "inner life" with AI, enabling AI-to-AI interactions that spark new human conversations.
• Competitive Landscape:
  • Startups can displace incumbents by introducing net-new interaction models and creative outlets that do not natively exist on incumbent social platforms.
• Product Mechanism:
  • Success depends on the core emotion of "being seen" and feeling connected.
  • Mechanisms to rapidly understand the user include ingestion of digital footprints, narration of life stories, and analysis of photo rolls.
• Future Application: AI agents will act as intermediaries to check in on friends and facilitate relationship building, moving beyond simple content creation.

David Haber (AI Business Model Reinforcement)

• Investment Thesis: AI strategies that reinforce business models and drive revenue outperform those focused solely on cost reduction and automation.
• Case Study: EVE (Plaintiff Law)
  • Model Structure: Operates on contingency fees (paid only upon winning), making AI's ability to take on more high-value cases critical.
  • Outcome: AI automates drafting and reasoning without eroding billable hours; instead, it enables attorneys to scale client intake and revenue.
  • Defensibility: Deep integration into the end-to-end workflow creates a proprietary dataset of case outcomes (intake to resolution) unavailable to general model labs.
• Case Study: Salient (Loan Servicing)
  • Technology: Deployed voice agents capable of speaking 50 languages, handling welcome calls, payment reminders, and regulatory compliance (UDAP).
  • Dual Value Proposition:
    • Cost Side: Reduces call center operational costs for lenders.
    • Revenue Side: Improves collection rates by delivering better outcomes for borrowers, reinforcing the lender's business model.
• Source of Competitive Advantage:
  • Compounding advantage arises from deep embedding within customer workflows to capture unique, non-public data (e.g., specific case outcomes).
  • This proprietary data feeds back into the AI to improve triage and strategy, creating a feedback loop that makes the platform smarter and more valuable with scale.
  • The market pull is strongest when AI directly correlates to increased customer revenue or success rates.