The first FDA approval for a CAR-T therapy in autoimmune disease is expected in the first half of 2026. The regulatory approvals that follow will determine which companies actually reach the 17.9 million RA patients.

Every health economist has run the same calculation: cost of drug A versus cost of drug B, measured in quality-adjusted life years, assessed over a 10-year horizon. That model is built for a world where nothing cures the disease.

The science is ready. The clinical data is published. The regulatory path is open. The factory is the problem.

For 30 years, the treatment model for most autoimmune diseases has rested on a single, uncomfortable truth: we cannot cure them. That just changed.

Daphne Koller’s insitro reduces drug development from 10-15 years to 3-6 years through ML-first architecture—computational prediction replacing 80% of wet lab experiments, achieving 80-90% Phase I success rates.

Dr. Ariella Heffernan-Marks built dual infrastructure enabling 48-hour vaccine manufacturing—how Ovum’s consumer product and research platform create bidirectional value addressing the $1 trillion gender health gap.

Alona Kharchenko achieved real-world humanoid robot deployment with €720K in grants while billion-dollar competitors remain years away—through hybrid intelligence architecture and strategic positioning in Germany’s care crisis.

Twenty years. One technical obsession. By March 2025, Andrea Thomaz’s Diligent Robotics had deployed robots that completed one million deliveries across dozens of hospitals—navigating elevator crowds, dodging stretchers in narrow corridors, and handing medication to nurses who were “so thankful” they posed for selfies with the machines.

This final installment synthesizes actionable frameworks, decision points, and strategic implications for operators, investors, and policymakers working in capital-intensive frontier technology domains.

At Sapien Fusion, credibility isn’t negotiable. Every claim in the Capital Efficiency Paradox series underwent systematic verification before publication. This document provides full transparency into our research methodology, source hierarchy, and verification protocols.

Most AI labs pursued incremental improvements on text and 2D image generation. Fei-Fei Li recognized something different during her 2017-2018 tenure as VP and Chief Scientist of AI/ML at Google Cloud: AI systems operated in a fundamentally two-dimensional paradigm, processing text tokens and image pixels without genuine spatial understanding.

In 2017, Joelle Pineau joined Meta to establish FAIR’s Montreal laboratory with 20-30 researchers. By 2025, when she left to become Chief AI Officer at Cohere, that research organization had released over 1,000 research artifacts—code, models, datasets—downloaded 2.7 billion times globally.