Pharma’s AI Shift: From Pilot Projects to Core Drug Discovery

At the Hilton London Kensington on March 9–10, 2026, senior scientists and executives from GSK, Sanofi, AbbVie and Insilico Medicine gathered for the 7th Annual AI in Drug Discovery Conference. Organised by SAE Media Group and billed as the largest AI in drug discovery event in the UK, the two-day meeting focused on how generative AI, large language models and autonomous laboratory agents are being deployed inside real pharmaceutical R&D pipelines rather than in isolated pilot studies.

The gathering matters beyond London because it reflects a structural shift in the global pharmaceutical industry. AI is no longer an experimental add-on; it is becoming a core R&D driver. According to event materials and market analysis, major drugmakers invested more than USD 5.8 billion in AI-focused partnerships between 2022 and 2024, and the first AI-designed drug candidates are now advancing through clinical development. For executives and specialists worldwide, the conference served as a barometer of how quickly the industry is moving from proof-of-concept to operational integration.

A strategic shift, not a technology demo

The speaker roster underscored the seniority now attached to AI roles inside large pharmaceutical organisations. Darren Green, former Head of Cheminformatics at GSK, and Ian Wall, Head of Molecular Design at GSK, were joined by Christoph Grebner, Senior Principal Scientist at Sanofi, and Esben Jannik Bjerrum, Principal Research Scientist II for AI/ML at AbbVie. Petrina Kamya, Global Head of AI Platforms at Insilico Medicine, brought the perspective of a company built around AI-native drug discovery.

Gold sponsor Schrödinger, a leader in physics-based computational drug discovery, was a key participant, reflecting the hybrid approach many companies now favour: combining machine learning with physics-based molecular simulation. The conference positioned itself around high-level networking and case studies from leading pharmaceutical firms, signalling that the industry has moved beyond broad promises and is now comparing operational results, implementation challenges and measurable R&D outcomes.

Generative AI, LLMs and autonomous labs dominate the agenda

The agenda concentrated on several themes that have rapidly moved from research papers into drug discovery workflows. These included:

• Generative AI for de novo drug design
• Large language models in biomedical research
• Autonomous agents in laboratories
• Target identification
• Data quality challenges

Generative AI for de novo drug design was a central topic, with presenters discussing how models can propose novel molecular structures with desired properties. Large language models in biomedical research were another focus, covering their use in mining scientific literature, structuring biological knowledge and supporting hypothesis generation.

Autonomous agents in laboratories also featured prominently, reflecting a push toward closed-loop systems in which AI designs experiments, interprets results and selects the next steps with minimal human intervention. Target identification and data quality challenges rounded out the programme. The event highlighted the growing impact of generative AI on R&D workflows and pointed to the clinical progress of the first AI-designed drug candidates as evidence that the technology is beginning to influence real therapeutic pipelines, not just early-stage discovery.

Market momentum and the economics of acceleration

The commercial backdrop is expanding quickly, though forecasts vary widely. One projection cited in event materials puts the global AI in drug discovery market at USD 7.94 billion by 2030. A separate analysis from Towards Healthcare forecasts a much larger expansion, from USD 24.51 billion in 2026 to USD 160.49 billion by 2035, a compound annual growth rate of 23.22%. The divergence reflects different definitions of the market and the speed at which AI capabilities are being embedded across discovery, preclinical and clinical workflows.

Underlying this growth are persistent economic pressures. Drug development costs average around USD 2.6 billion per drug, and AI has the potential to reduce preclinical timelines by 30–40%, according to market research cited around the event. Those savings are not theoretical: shorter preclinical phases can mean earlier clinical readouts, lower capital intensity and faster decision-making on which candidates to advance or kill. For an industry facing escalating development costs and evolving regulatory expectations, that acceleration is a strategic imperative rather than a marginal efficiency gain.

Data quality and regulatory pathways remain the hard part

Despite the momentum, the conference did not shy away from the obstacles. Data quality was repeatedly identified as a limiting factor. AI models are only as reliable as the experimental and clinical data they are trained on, and many pharmaceutical datasets remain fragmented, inconsistent or locked in proprietary silos. Speakers and attendees discussed practical approaches to curating training data, standardising assay outputs and ensuring that model predictions can be traced back to reproducible experimental evidence.

Regulatory pathways for AI-generated therapeutics are still evolving. While the first AI-designed candidates have entered clinical trials, regulators in major markets have not yet established fully harmonised frameworks for assessing how such molecules are discovered, validated and manufactured. The conference’s emphasis on real-world case studies and networking reflected the industry’s need to share best practices for integrating AI, managing data quality and navigating those regulatory uncertainties. For global pharmaceutical leaders, the message was clear: the technology is maturing faster than the governance around it, and companies that solve data and regulatory integration early will be better positioned to extract value from AI investments.

Looking ahead, the 7th Annual AI in Drug Discovery Conference suggests that the next phase of AI adoption in pharma will be defined less by model breakthroughs and more by operational discipline. The tools for generative design, language-based knowledge extraction and autonomous experimentation are already in place at major companies. The differentiator will be how effectively organisations connect those tools to clean data, robust validation and regulatory-ready workflows. As the market projections indicate—ranging from USD 7.94 billion by 2030 to USD 160.49 billion by 2035—the financial stakes are rising quickly, but the path from promising algorithm to approved medicine still runs through the same rigorous scientific and regulatory requirements that have always governed drug development.