When AI Meets Biotech, Geography Still Matters: Inside the France–Taiwan Debate on Drug Discovery, Capital, and Scale

The global race to dominate the next industrial era is shifting from pure silicon to the fusion of artificial intelligence and biotechnology. As AI reshapes drug discovery, industry leaders increasingly agree that the “next TSMC” will not be built on algorithms alone.

The France–Taiwan Executive Dialogue on AI & Biotech Innovation, co-hosted by InFocus Therapeutics and Business France Taiwan, convened senior executives from pharma, venture capital, and government for a high-level working session on unresolved questions facing the sector: Where does AI genuinely add value? What slows progress once algorithms leave the lab? And how can countries collaborate without losing strategic control over their innovation ecosystems?

A Sector Searching for Its “Next TSMC”

Opening the event, Franck Paris, Director of the French Office in Taipei, set a provocative tone by linking biotech’s future to the lessons of the semiconductor industry. “We always speculate about what will be the next TSMC or the next NVIDIA,” he said. “Many are convinced—and rightly so—that the next one will be in biotech.”

But unlike semiconductors, where Taiwan dominates through manufacturing scale and speed, biotech innovation depends on long timelines, regulatory endurance, and a staggering tolerance for failure. Paris emphasized that France and Taiwan bring complementary strengths: France’s decades-long investment in fundamental research and hospital data systems, and Taiwan’s unmatched computing infrastructure and capital markets. The challenge, he argued, is not technology availability but the integration of these two distinct worlds.

Franck Paris, Director of the French Office in Taipei (featured), opened the France–Taiwan Executive Dialogue by highlighting AI-driven biotech as a strategic frontier for bilateral collaboration, emphasizing the need to align research ecosystems, capital, and industrial scale to turn long-term public investment in innovation into globally competitive drug discovery outcomes. Image: GeneOnline

Efficiency vs. Miracle Cures: The Reality of the Lab

For Jérémy Besnard, former co-founder and chief designer of Exscientia, the industry has already moved past the idea that AI alone can “solve” biology. Having helped scale Exscientia from a three-person startup to a $3 billion Nasdaq-listed company, Besnard described a persistent gap between investor expectations and biological reality.
“AI helps people be more efficient,” Besnard said. “What it really changes is how drugs are discovered—not whether biology works.” He highlighted the “hard work” of educating investors who often expect rapid software-like monetization, whereas life-science returns may take a decade of patient platform-building before a single drug reaches the clinic.

This tension reflects a broader recalibration across the AI-biotech sector. After several years of exuberant claims that algorithms could dramatically compress development timelines, many companies are now repositioning AI as an enabling infrastructure rather than a replacement for experimental biology. In practice, this means earlier go/no-go decisions, narrower target selection, and reduced downstream attrition—incremental gains that compound over time, but rarely translate into immediate clinical breakthroughs.

This sentiment was reinforced by Jean-François Mirjolet, Chief R&D Head of InFocus, who focused on the limits of machine learning. 

“AI can help with precision and speed, allowing us to screen more compounds virtually,” Mirjolet said. “But what AI cannot do is create data. Our field publishes success; failure is often hidden.” He warned that without registering “negative data”—the compounds that failed—AI tools risk reinforcing existing biases.

The challenge of incomplete datasets remains one of the sector’s most underappreciated bottlenecks. While regulatory agencies and journals increasingly recognize the value of negative results, structural incentives still reward novelty over transparency. Until failure becomes as systematically recorded as success, AI models risk optimizing against an artificially optimistic version of biology—limiting their ability to predict clinical outcomes reliably.

The Bigger Perspective: Agile Startups vs. Global Giants

From the perspective of large pharmaceutical companies, AI adoption is a matter of agility versus scale. Nicolas Martin, Portfolio Solution Lead at Roche, described a decentralized reality where AI technology “pops up” in different teams at different speeds across a global organization. While smaller “AI-native” companies like InFocus or Exscientia are built end-to-end on technology, Martin noted that the giants are catching up. 

“We gain efficiency in the early stages,” he said, “but biology always catches up in the clinic.” He likened the current wave of AI enthusiasm to earlier technology cycles in genomics—periods marked by high expectations followed by a return to the fundamental reality of human trials.

Across large pharma, AI integration increasingly takes the form of targeted deployment rather than wholesale transformation. Rather than replacing existing discovery engines, companies are embedding machine learning into discrete stages—hit identification, patient stratification, biomarker discovery—where marginal gains can meaningfully reduce cost or failure risk. The result is slower but more controlled adoption, shaped by regulatory accountability and the need to validate tools across diverse therapeutic areas.

Roche Portfolio Solution Lead Nicolas Martin (left) and Jean-François Mirjolet, former Head of Research at Inventiva (right), discussed where AI delivers tangible gains in drug discovery—improving speed and precision in early-stage screening—while underscoring that clinical success remains constrained by biology, data completeness, and regulatory limits. Image: GeneOnline

Capital with a Conscience: The Taiwan Difference

A recurring theme was the nature of the capital required to fuel this “bioconvergence.” Karen Yu, President of the Industrial Technology Investment Corporation (ITIC), contrasted biotech with Taiwan’s traditional ICT sector.

“Semiconductors have short life cycles and intense competition,” Yu said. “Life sciences depend on patents and long regulatory endurance.” At ITIC, the investment strategy has shifted to look beyond financial returns toward “strategic value.” Taiwan’s advantage lies in its ability to integrate hardware, computation, and industrial scale—capabilities increasingly relevant as drug discovery becomes more data-intensive. “Taiwan can help solve the data and computation challenge that life science is currently facing,” Yu added.

This approach reflects a growing recognition that biotech returns cannot be evaluated through the same metrics used for electronics or software startups. In drug development, value creation often occurs years before revenue, embedded in validated platforms, clinical optionality, and regulatory progress. Taiwan’s capital markets—historically more tolerant of long development arcs in life sciences—position the island as a complementary funding environment for AI-biotech companies navigating early risk.

Sovereignty without Isolation: Building the Corridor

As geopolitical concerns shape technology policy, the panel addressed the tension between national sovereignty and global collaboration. The emerging consensus was one of “open sovereignty”: anchoring core technologies locally while remaining open to global talent and clinical validation.

This model has gained traction as countries seek to secure strategic technologies without isolating themselves from global innovation flows. In biotech, where clinical trials, patient diversity, and regulatory pathways span borders by necessity, rigid national silos risk slowing progress rather than protecting it. The emphasis has shifted toward retaining control over core IP and data while enabling collaboration at the level of talent, capital, and execution.

Emily Fang, Founder and CEO of InFocus Therapeutics, framed this cross-border collaboration as an alliance rather than a compromise.

“Money enables us to do things, technology enables us to do things, but goodwill is also a form of capital,” Fang said. She described her role as a translator between technologists, scientists, and investors across different cultures. For Fang, success is measured by the creation of an innovation “corridor” between hubs like Paris and Taipei. “The talent and know-how need to flow,” she said. “When people are diagnosed with a serious disease, they want to know someone is working on it. That hope is the ultimate North star.”

From left to right, Jérémy Besnard, former Co-Founder and Chief Designer of Exscientia; Karen Yu, President of Industrial Technology Investment Corporation (ITIC); Emily Fang, Founder and CEO of InFocus Therapeutics; and moderator Volker Heistermann, Managing Director of MOSAIC Venture Lab, examined how AI-driven drug discovery is reshaping investment expectations and cross-border collaboration, particularly at the intersection of semiconductor-enabled computation and biotech innovation in France and Taiwan.Image: GeneOnline

From Promise to Proof: Looking Ahead as AI Biotech Moves From Narrative to Execution

The France–Taiwan dialogue made clear that the next phase of AI-driven biotech will be defined less by algorithmic novelty than by execution across capital, data, regulation, and geography. As AI becomes embedded in drug discovery workflows, its impact will be judged not by speed alone, but by whether it reduces downstream failure, aligns with long clinical timelines, and translates across borders. The winners are unlikely to emerge from isolated breakthroughs, but from ecosystems able to integrate computation, biological insight, and patient capital at scale.

Key considerations going forward:

• AI’s real value will show up in fewer late-stage failures, not headline-making “breakthroughs,” as early stop decisions and better target prioritization reshape R&D efficiency.
• Data completeness—especially negative and failed results—will become a differentiator, separating robust discovery platforms from overfit models trained on success alone.
• Capital patience will matter as much as technical capability, favoring investors and markets that understand platform economics and long clinical arcs over software-style return expectations.
• Cross-border “corridors” will outperform single hubs, as regions that link research depth, compute power, capital, and regulatory pathways gain structural advantage.
• Execution—not algorithms—will determine leadership, as AI moves from promise to infrastructure in the next generation of biopharma innovation.

What emerged from the France–Taiwan Executive Dialogue was not a vision of AI as a silver bullet, but as a structural tool—one that reshapes how decisions are made, risks are managed, and ecosystems are built. The discussion underscored that drug discovery remains a human, biological, and regulatory endeavor, even as computation accelerates its early stages. The real opportunity now lies in aligning incentives across science, capital, and policy, and in building cross-border frameworks capable of sustaining innovation through long and uncertain development cycles. In that sense, the next industrial leaders in biotech will not be defined by who adopts AI first, but by who integrates it most responsibly—and at scale.

Author

Bernice Lottering