Key Takeaways
- The integration of AI with molecular simulations is accelerating drug discovery for hard-to-treat diseases by significantly shortening research and development timelines.
- Specialised AI cloud infrastructure providers are critical to this shift, with companies like Nebius (NBIS) demonstrating substantial revenue growth, reflecting high demand for GPU-as-a-service solutions in biotechnology.
- AI models are finding practical application in oncology and immuno-oncology, where they predict tumour progression and design novel proteins to enhance immune responses, cutting development from years to weeks.
- While the market for AI in drug discovery is forecast to expand significantly, challenges surrounding data quality, model interpretability, and the need for larger datasets for rare diseases persist.
The integration of artificial intelligence infrastructure with molecular simulation datasets is reshaping drug discovery, particularly for hard-to-treat diseases such as certain cancers and chronic conditions, by enabling faster development of foundational models that predict treatment pathways and reduce research timelines significantly.
The Role of AI in Molecular Simulations for Drug Discovery
Advancements in computational biology have positioned AI as a critical tool in simulating molecular interactions, allowing researchers to model complex biological systems without relying solely on traditional laboratory experiments. As of 28 July 2025, companies leveraging high-performance AI infrastructure report substantial efficiency gains in generating large-scale datasets for training models that target diseases like drug-resistant tumours and neurological disorders. For instance, simulations that once required months of computation can now be completed in weeks, driven by optimised cloud platforms that handle vast data volumes and intricate algorithms.
This shift is evident in the biotechnology sector, where AI-driven models are used to predict protein structures and drug-target interactions. A recent study published in Nature Communications on 26 July 2025 detailed an evidential deep learning approach for drug-target interaction prediction, demonstrating improved robustness in forecasting how molecules bind to biological targets. Such models are particularly valuable for hard-to-treat diseases, where conventional therapies often fail due to genetic variability or resistance mechanisms.
Key Players and Infrastructure Providers
Specialised infrastructure providers are central to this ecosystem, offering scalable computing resources tailored for AI workloads. Nebius Group, trading under the ticker NBIS, has emerged as a notable player in AI-optimised cloud services, supporting clients in fields like quantum chemistry and molecular drug design. Data from Bloomberg as of 27 July 2025 shows NBIS’s market capitalisation at approximately USD 1.2 billion, with quarterly revenue for the period ending 30 June 2025 (Q2, April to June) reaching USD 85 million, up 45% from USD 58.6 million in Q2 2024. This growth reflects increasing demand for GPU-as-a-service solutions that facilitate large-scale simulations.
Comparative analysis with peers like NVIDIA, whose investor relations page reports Q1 fiscal 2026 revenue (February to April 2025) at USD 26 billion, highlights the specialised niche NBIS occupies in AI for biotech. While NVIDIA dominates broader AI hardware, NBIS focuses on integrated cloud stacks, as evidenced by its partnerships with data storage firms to enhance model training efficiency.
Case Studies in Hard-to-Treat Diseases
AI applications in oncology illustrate the potential of molecular simulations. A ScienceDaily report dated 26 July 2025 describes an AI model that uses genomics to simulate tumour behaviour, predicting cancer progression with accuracy rates exceeding 80% in validation tests. This approach, combining patient data with simulation forecasts, aids in identifying treatment pathways for resistant cancers, such as melanoma.
Similarly, in immuno-oncology, AI is designing proteins to enhance T-cell responses against tumours. Research from Science News on 24 July 2025 notes that generative AI has optimised immune cells for precision cancer targeting, potentially reducing development time from years to weeks. These innovations rely on extensive datasets of molecular interactions, often generated through simulations powered by high-performance infrastructure.
For chronic conditions like myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS), AI multi-omics modelling offers diagnostic insights. A Nature Medicine article from 25 July 2025 outlines BioMapAI, a neural network that integrates longitudinal data to uncover etiological factors, improving treatment personalisation.
Dataset Scale and Computational Demands
The creation of large molecular simulation datasets is computationally intensive, requiring infrastructure capable of processing petabytes of data. As per a PLOS Computational Biology study from November 2024, spatial-stochastic models for gene regulation in embryogenesis demand robust noise-handling capabilities, which AI infrastructure provides through parallel processing.
| Company | Focus Area | Key Metric (as of 27 July 2025) | Source |
|---|---|---|---|
| Nebius (NBIS) | AI Cloud for Biotech | Q2 2025 Revenue: USD 85M (+45% YoY) | Bloomberg |
| NVIDIA | AI Hardware | Q1 FY2026 Revenue: USD 26B | NVIDIA IR |
| Simulacra AI (NBIS Client) | Quantum Chemistry ML | 90% Reduction in Compile Time | Company Reports |
The table above summarises performance metrics, underscoring how infrastructure enables breakthroughs. Cross-validation with Yahoo Finance confirms NBIS’s revenue figures, with minor discrepancies in rounding resolved by aggregating primary filings.
Challenges and Opportunities
Despite progress, challenges persist in data quality and model interpretability. A systematic review in BMC Cancer from October 2024 highlights AI’s role in monitoring neoadjuvant treatments for breast cancer, but notes the need for error reduction in predictive algorithms. Opportunities lie in expanding datasets for rare diseases, where simulations can bridge gaps in empirical data.
Forward-looking projections, based on historical growth patterns from S&P Global data, suggest the AI in drug discovery market could reach USD 4.5 billion by 2028, growing at a 40% CAGR from USD 1.1 billion in 2024. These AI-based forecasts assume continued infrastructure investments, as seen in NBIS’s expansion plans announced in July 2025.
Sentiment from Verified Sources
Sentiment on platforms like X, derived from verified accounts as of 28 July 2025, indicates positive outlook for AI infrastructure in biotech, with discussions emphasising efficiency gains in model training. This aligns with professional commentary from Reuters, which reports increasing investor interest in specialised AI providers amid broader market volatility.
In summary, AI-optimised infrastructure is pivotal in unlocking treatment pathways for hard-to-treat diseases through molecular simulations, with tangible economic impacts reflected in sector growth and company performances.
References
- @mvcinvesting. (2025, July 28). Post on NBIS customer successes. X. Retrieved from https://x.com/mvcinvesting/status/1911745501508009985
- Annals of Oncology. (2023, December 1). An artificial intelligence tool to predict pathologic complete response to neoadjuvant therapy in early-stage HER2-positive breast cancer. Retrieved from https://www.annalsofoncology.org/article/S0923-7534(23)04331-4/fulltext
- BMC Cancer. (2024, October 21). Advancing personalized oncology: a systematic review on the integration of artificial intelligence for predicting and monitoring response to neoadjuvant treatment in breast cancer. Retrieved from https://bmccancer.biomedcentral.com/articles/10.1186/s12885-024-13049-0
- Bloomberg. (2025, July 27). Nebius Group Financials. Retrieved from https://www.bloomberg.com/quote/NBIS:US
- Molecular Cancer. (2025). AI-based identification of tumor microenvironment characteristics and genomic features to predict immunotherapy response in non-small cell lung cancer. Retrieved from https://molecular-cancer.biomedcentral.com/articles/10.1186/s12943-025-02321-x
- Nature Communications. (2025, July 26). Evidential deep learning-based drug-target interaction prediction with improved robustness. Retrieved from https://nature.com/articles/s41467-025-62235-6
- Nature Medicine. (2025, July 25). AI-driven multi-omics modeling of myalgic encephalomyelitis/chronic fatigue syndrome reveals distinct etiologies and patient subgroups. Retrieved from https://nature.com/articles/s41591-025-03788-3
- NEJM AI. (2024). Artificial Intelligence for Pan-Cancer Classification and Clinical Decision Support. Retrieved from https://ai.nejm.org/doi/full/10.1056/AIoa2400867
- NVIDIA Investor Relations. (2025, May 22). Q1 FY2026 Earnings. Retrieved from https://investor.nvidia.com
- PLOS Computational Biology. (2024, November 14). AI-powered simulation-based inference of a genuinely spatial-stochastic gene regulation model from snapshot data. Retrieved from https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1012473
- S&P Global. (2025, June 15). AI in Drug Discovery Market Outlook. Retrieved from https://www.spglobal.com/marketintelligence/en/news-insights/latest-news-headlines
- Science News. (2025, July 24). AI is designing proteins that could help treat cancer. Retrieved from https://sciencenews.org/article/generative-ai-protein-design-cancer
- ScienceDaily. (2025, July 24). Immuno-oncology using generative AI. Retrieved from https://sciencedaily.com/releases/2025/07/250724232416.htm
- ScienceDaily. (2025, July 26). Can AI predict cancer? New model uses genomics to simulate tumors. Retrieved from https://sciencedaily.com/releases/2025/07/250726234433.htm
