The pharmaceutical industry is moving decisively beyond experimental AI pilots into large-scale, production-ready drug discovery systems. Three major developments announced this week reveal how artificial intelligence is reshaping the speed and economics of finding new medicines, from identifying novel disease targets to screening billions of molecular candidates at a fraction of previous costs (Sources 1, 2, 3).

What Is Causal AI, and Why Does It Matter for Neurological Diseases?

Ono Pharmaceutical, a major Japanese drugmaker, announced a partnership with Aitia, a Cambridge-based biotech company, to leverage a new approach called causal AI for discovering treatments in neurology. Unlike traditional machine learning, which identifies patterns and correlations in data, causal AI goes deeper: it infers why things happen, not just that they happen together.

The partnership centers on Aitia's REFS technology, which builds what the company calls "Gemini Digital Twins" (GDTs). These are virtual patient models that reflect real disease characteristics and allow researchers to identify true therapeutic targets rather than false correlations. For neurological diseases, where disease mechanisms involve complex interactions among genetic, developmental, neural network, immune, and environmental factors, this distinction is critical.

"Through this collaboration, we are delighted to leverage cutting-edge REFS technology to identify true therapeutic targets in neurological diseases, where disease mechanisms are complex and not yet fully understood. We believe this will accelerate the development of innovative medicines," said Seishi Katsumata, Corporate Officer and Executive Vice President of Discovery and Research at Ono.

Seishi Katsumata, Corporate Officer and Executive Vice President of Discovery and Research, Ono Pharmaceutical

Under the agreement, Ono gains exclusive worldwide rights to research, develop, and commercialize drug candidates targeting any therapeutic targets identified by Aitia's technology. This structure reflects growing confidence in AI-driven target discovery as a legitimate foundation for billion-dollar drug development programs.

How Can AI Screen Billions of Drug Molecules in Hours Instead of Months?

A separate breakthrough published in Nature Biotechnology demonstrates the computational revolution underway in virtual drug screening. Researchers from St. Jude Children's Research Hospital, University of Pavia, Dana Farber Cancer Institute, and Harvard Medical School unveiled AdaptiveFlow, an open-source platform that can virtually screen 69 billion drug-like molecules with a 1,000-fold reduction in computational costs compared to existing methods.

The platform's innovation lies in its ability to scale efficiently across massive computing infrastructure. AdaptiveFlow demonstrated linear scaling up to 5.6 million virtual central processing units (CPUs), a new benchmark for cloud-based drug discovery. Most software loses efficiency as computing power increases, but AdaptiveFlow maintains perfect linear scaling, meaning researchers can add more computing power without wasting resources.

At the core of AdaptiveFlow is an 18-dimensional grid where each dimension represents a specific molecular property, such as molecular weight. This framework allows researchers to prioritize chemically diverse molecules and guide rational selection of promising library subsets for deeper screening. A machine-learning classification model then identifies the most promising candidates, which are virtually screened against protein targets using over 1,500 supported docking protocols.

"With AdaptiveFlow, the scaling behavior is perfectly linear, even with millions of CPUs. That is special," explained Christoph Gorgulla, PhD, Center of Excellence for Data-Driven Discovery at St. Jude Department of Structural Biology.

Christoph Gorgulla, PhD, Center of Excellence for Data-Driven Discovery, St. Jude Department of Structural Biology

To prove the platform's effectiveness, the team identified potent inhibitors for two cancer targets: poly(ADP-ribose) polymerase 1 (PARP1), an established target with approved drugs on the market, and ferroptosis suppressor protein 1 (FSP1), an emerging target involved in cell death and cancer cell survival. The platform successfully identified inhibitors with binding strength comparable to pharmaceutically relevant standards, even for the more challenging FSP1 target.

Critically, AdaptiveFlow is open-source and available for free download, democratizing access to ultra-large virtual screening capabilities that were previously prohibitively expensive.

Why Are Pharma Companies Investing So Heavily in AI Right Now?

A survey of pharmaceutical supply chain professionals reveals the industry's confidence in AI as a strategic priority. Research from LogiPharma found that 96 percent of respondents ranked artificial intelligence and machine learning among their top investment priorities, making AI the most frequently cited strategic focus across the sector.

The survey identified specific application areas where pharma leaders expect AI to deliver the greatest value:

• Demand Planning and Forecasting: AI models predict which drugs will be needed in which quantities, reducing waste and ensuring availability.
• Inventory Optimization: Machine learning algorithms balance stock levels across manufacturing and distribution networks to minimize costs while maintaining supply reliability.
• Logistics Orchestration: AI coordinates complex supply chains involving multiple suppliers, manufacturers, and distribution centers to reduce delays and costs.

However, the survey also revealed significant barriers to broader adoption. Regulatory uncertainty and compliance concerns were identified as the single biggest obstacle, with many organizations uncertain about how to implement AI responsibly within existing regulatory frameworks. Additionally, more than half of respondents remain uncertain about AI's ability to meaningfully improve disruption prediction and mitigation, suggesting that while enthusiasm is high, confidence in specific outcomes remains mixed.

How to Evaluate AI Drug Discovery Partnerships for Your Organization

• Assess Technology Differentiation: Determine whether the AI platform uses novel approaches like causal inference or simply applies conventional machine learning. Causal AI, as demonstrated by Aitia's partnership with Ono, offers advantages in identifying true disease mechanisms rather than spurious correlations.
• Verify Computational Efficiency: Evaluate whether the platform can scale linearly across large computing infrastructure without loss of efficiency. AdaptiveFlow's ability to maintain perfect linear scaling across millions of CPUs demonstrates this capability and translates to lower costs.
• Review Validation Evidence: Examine whether the platform has been tested on both well-characterized targets with existing drugs and challenging emerging targets. Success on both types indicates robustness and real-world applicability.
• Understand Governance and Compliance: Clarify how the partnership addresses regulatory requirements, data privacy, and intellectual property ownership. Ono's exclusive worldwide rights to commercialize targets identified by Aitia exemplifies clear contractual structure.

The convergence of these three developments signals a maturation of AI in drug discovery. Causal AI platforms are moving from research curiosities into production partnerships with major pharmaceutical companies. Computational breakthroughs are making ultra-large virtual screening economically feasible for routine use. And industry-wide surveys confirm that AI investment is no longer a speculative bet but a core strategic priority for 96 percent of pharma leaders (Sources 1, 2, 3).

The remaining challenge is not technological but organizational: pharma companies must build governance frameworks, compliance processes, and internal expertise to deploy these tools responsibly and extract maximum value. The companies that solve this puzzle first will likely gain significant competitive advantages in identifying and developing the next generation of breakthrough medicines.