In the past decade, gene therapies have moved from science fiction to clinical reality. Treatments like Hemgenix, recently approved for haemophilia B, can effectively cure devastating genetic conditions with a single infusion. For patients, the results are life-changing. Yet the cost is staggering: Hemgenix is priced at $3.5 million per dose, making it the most expensive drug in the world. Other therapies, such as Zynteglo for beta thalassaemia and Zolgensma for spinal muscular atrophy, carry similar multimillion-dollar price tags.

The problem is clear: if gene therapies remain this expensive, healthcare systems and insurers will struggle to provide access, even in wealthy nations. Bluebird Bio, for example, withdrew Zynteglo from the European market in 2021 after governments refused to reimburse its $1.8 million cost . For rarer conditions, high per-patient costs may still be justified, but as gene therapies expand to larger patient groups—such as those with sickle cell disease, which affects more than 100,000 Americans—health budgets could buckle.

So how do we square the circle—delivering transformative cures at prices that health systems can bear? The answer lies not only in manufacturing scale but also in artificial intelligence (AI) and data innovation.

Why AI Matters for Gene Therapy Costs

The development of a single gene therapy can cost up to $5 billion, nearly five times that of a conventional drug . Much of this expense comes from the long, uncertain process of discovery, trial design, and regulatory approval. AI has the potential to transform this pipeline by:

1. Accelerating target discovery: Machine learning models can analyze genomic and clinical datasets to identify which mutations are most actionable and predict therapeutic outcomes.

2. Optimizing clinical trials: AI can match patients more effectively, simulate outcomes, and flag likely non-responders before trials even begin—reducing failure rates.

3. Improving manufacturing efficiency: Data-driven process control can cut the costs of producing complex viral vectors and other delivery systems.

   
   

By reducing uncertainty and waste at every stage, AI can make gene therapies faster to develop, safer, and ultimately more affordable.

Why Data Diversity is the Missing Link

But AI is only as good as the data it trains on. Currently, the vast majority of clinical and genomic datasets come from populations of European ancestry, even though they represent less than 15% of the global population. This imbalance has real consequences:

• Drug efficacy and safety vary across populations. Statins, one of the most prescribed drugs worldwide, show different side-effect profiles in East Asian populations compared to Europeans. If datasets ignore such variation, guidelines may be inappropriate or even harmful.

• Diseases of underrepresented groups get less attention. Sickle cell disease, which predominantly affects people of African descent, has historically been underfunded in research and clinical trials despite its high burden.

  
  

For AI to reduce costs and improve gene therapy development, it must be trained on authentic, compliant, and highly diverse clinical and genomic data. Without this, predictive models risk being biased, ineffective, or even unsafe when applied globally.

Building a Future of Accessible Genomic Medicine

To make gene therapies truly transformative, we must invest in three pillars:

• Data infrastructure: Governments and institutions must collaborate on global platforms that securely share diverse, de-identified patient data.

• AI innovation: Companies and researchers need access to these datasets to build robust models that can streamline discovery and trials.

• Policy frameworks: Strong governance must ensure that data use is ethical, compliant, and equitable.

The promise of gene therapies is immense—but so is the risk of financial unsustainability . By unlocking the combined power of AI and diverse clinical data, we can move beyond today’s eye-watering price tags and towards a future where lifesaving cures are available to all who need them.

References

Financial Times: Gene therapies may cure disease but can we afford them? (Dec 15, 2022)

JAMA Cardiology: Disparities in Statin Use Among US Adults (2023)

European Heart Journal – Cardiovascular Pharmacotherapy: Ethnic differences in statin sensitivity (2022)

Costa, E. et al. (2025). Globalization in clinical drug development for sickle cell disease. American Journal of Hematology, 100(1), 4-9. DOI: 10.1002/ajh.27525