8 December 2025 · 7 min
You're Virtualising Everything Except Your Drug
Why 79% of pharmaceutical companies virtualise their bioreactors and production lines but leave the most critical variable - the drug formulation itself - trapped in an analogue world of trial-and-error experimentation.
The Paradox of Pharmaceutical Digitisation
Pharma has gone all-in on digital twins. The global digital twin market in pharma hit $1.3 billion in 2025 and is on track to reach $8.5 billion by 2032, growing at a 30.2% CAGR. 79% of pharmaceutical firms already use digital twins for design precision. 63% of production lines are expected to adopt them by 2028.
But there is a glaring blind spot. Companies virtualise bioreactors, fermenters, and production equipment down to the sensor level while (the drug formulation itself) stays locked in empirical trial-and-error. Only 17% of pharmaceutical companies have facility-wide digital twins, and formulation development remains almost entirely analogue.
This is not just a missed opportunity. 70-90% of drug candidates exhibit poor solubility. Each drug presents over 3.6 million possible formulations. Without virtualised formulation development, the industry burns billions annually on inefficiencies that delay therapies from reaching patients.
Where Digital Twins Thrive And Where They Don't
Digital twins work exceptionally well for equipment monitoring, process control, and manufacturing optimisation. Bioreactor twins track temperature, pH, dissolved oxygen, and nutrient levels in real time, flagging deviations before they compromise product quality. Production line twins drive process adjustments and predictive maintenance, cutting downtime and waste.
The results speak for themselves: 35% faster time-to-market, 43% higher yields, and 18-28% cost reductions in manufacturing operations. These numbers explain why digital twin investment keeps accelerating.
Formulation development tells a different story. The process of combining active pharmaceutical ingredients with excipients to create stable, bioavailable, manufacturable drug products still relies on traditional wet-lab approaches. Scientists run physical experiments one formulation at a time, testing combinations that computational methods could predict in seconds.
Controlling variability allows us to improve quality and make product 'right first time' every time.
Source: GSK, on digital transformation in manufacturing
Why Formulation Remains Analogue: The Root Causes
Four interconnected challenges explain why formulation virtualisation has lagged behind equipment monitoring.
- Molecular Complexity: Equipment has well-characterised physical parameters. Molecular interactions do not. Quantum mechanical effects, thermodynamic complexity, and emergent behaviours across atomic-to-macroscopic scales make API-excipient interactions far harder to model than bioreactor temperature curves.
- Model Interoperability Barriers: Formulation digital twins demand integration across molecular dynamics, thermodynamic models, process simulations, and empirical correlations. Building a unified platform that connects these disparate modelling paradigms remains a hard engineering problem.
- Data Gaps: Equipment generates continuous, high-volume sensor data -- ideal training material. Formulation development produces sparse, expensive data points with high measurement variability. Standard ML approaches choke on these limited datasets.
Regulatory Uncertainty: Regulators accept PAT and equipment monitoring. Computational formulation development faces open questions around validation, transparency, and the regulatory pathway for AI-driven formulation decisions.
The Solubility Crisis Multiplier
The disconnect compounds an already severe challenge: 70-90% of drug candidates in development pipelines exhibit poor aqueous solubility. These molecules need advanced formulation strategies -- amorphous solid dispersions, lipid-based systems, nanoparticle technologies -- just to achieve adequate bioavailability.
For each poorly soluble compound, the design space detonates. Excipient selection, polymer ratios, processing parameters, and manufacturing conditions generate millions of combinations. Over 3.6 million potential formulations per drug, no computational guidance, and teams default to exhaustive screening campaigns that burn through years of work and scarce API supplies.
Formulation digital twins would deliver the most value here -- and this is exactly where the industry has been slowest to adopt computational methods. The hardest formulation problems are still the most manual.
Digital Twin Adoption: Equipment vs. Formulation
| Aspect | Equipment Digital Twins | Formulation Digital Twins |
|---|---|---|
| Adoption Rate | 79% of pharma firms | <17% facility-wide |
| Data Availability | Continuous sensor streams | Sparse experimental data |
| Model Maturity | Well-established physics | Emerging ML/AI approaches |
| Regulatory Clarity | PAT framework established | Evolving guidance |
| ROI Visibility | Direct cost reduction | Accelerated development |
| Opportunity Gap | Mature, optimised | Massive untapped potential |
Computational modeling and simulation play a critical role in organizing diverse data sets and integrating knowledge across development stages.
Source: FDA, on the role of computational approaches in drug development
Computational Formulation Design: Bridging the Gap
Computational formulation design offers a direct path forward. Rooted in the FDA's Quality by Design framework, it extends digital twin principles to the formulation itself, building virtual representations of drug products that predict critical quality attributes before any physical experiment runs.
The prediction accuracy is already there. ML models for formulation prediction now routinely exceed R² of 0.96, generating results in seconds instead of the months that experimental campaigns require. These platforms evaluate millions of formulation combinations computationally and surface the strongest candidates for targeted experimental validation.
The regulatory landscape is moving in the same direction. The FDA received over 500 AI-related submissions between 2016 and 2023. The agency has stated explicitly that AI models could "more quickly identify optimal processing parameters or scale-up processes, reducing development time and waste."
35%
Faster Time-to-Market for Adopters
43%
Higher Yields Reported
18-28%
Cost Reductions Achieved
AI-Powered Formulation Platforms
Platforms including Schrodinger Formulation ML, FormulationAI, and ExPreSo are proving that computational formulation design works at production scale.
R² > 0.96 Prediction Accuracy
Regulatory Momentum
The FDA logged 500+ AI-related submissions between 2016 and 2023 -- a clear signal that computational methods are gaining regulatory traction in pharma development.
500+ AI Submissions to FDA
By using AI/ML, scientists can streamline the formulation process, effectively narrowing down the design space from millions of possibilities to a tractable set of candidates.
Source: PharmTech, on AI in formulation development
The Digital Twin Opportunity
30.2%
Market CAGR to 2032
70-90%
Poorly Soluble Candidates
3.6M+
Formulations Per Drug
Seconds
AI Prediction Time
The Strategic Imperative
The formulation gap is the single largest untapped opportunity in pharmaceutical digitisation. Equipment and manufacturing virtualisation have proved their worth. The formulation itself is next.
The technical barriers are gone. AI/ML platforms deliver R² > 0.96 prediction accuracy. Regulators are accepting computational approaches. Early adopters have demonstrated clear ROI. What remains is organisational commitment and strategic capital allocation.
63% of production lines will use digital twins by 2028. The market is growing at 30.2% CAGR. The question is not whether formulation digital twins become standard practice -- it is which organisations move first and lock in the competitive advantage.
First movers will compress development timelines, cut costs, raise success rates, and get therapies to patients faster. A digital twin strategy that stops at the equipment is only half a strategy.

