In the high-stakes world of pharmaceutical development, where the average cost to bring a new therapy to market has ballooned to an estimated $2.67 billion, efficiency is no longer a luxury—it is a survival mandate. At the intersection of artificial intelligence and immunology, a New York-based startup, Immunai, has emerged as a critical partner for global pharmaceutical giants. By building a foundational model of the human immune system, Immunai is helping to solve the industry’s most stubborn "plumbing" problems, prompting AstraZeneca to extend their high-impact collaboration for the third time.
The latest expansion, announced recently, positions Immunai to receive up to $37.5 million through 2026 and 2027. This financial commitment underscores the deepening reliance that multinational corporations are placing on specialized AI infrastructure to de-risk clinical trials and accelerate the delivery of life-saving medicines.
A Chronology of Collaboration: From Pandemic Roots to Pipeline Integration
The relationship between Immunai and AstraZeneca is not a recent convenience but a long-standing strategic alliance that predates their formal public engagements. According to Immunai CEO Noam Solomon, the roots of the partnership reach back roughly five years to the height of the global pandemic, a period that forced the entire pharmaceutical industry to rethink its approach to immunology and rapid data processing.
- Late 2022: The official collaboration commences, initially focusing on oncology—a cornerstone of AstraZeneca’s therapeutic portfolio.
- October 2025: The partnership pivots beyond oncology, signaling a significant expansion into Inflammatory Bowel Disease (IBD). This move marked a shift in Immunai’s utility, proving that their immune-system foundation model was applicable across diverse inflammatory conditions.
- Present Day (2026): The third major expansion is inked. The scope now encompasses not only oncology and IBD but also cardiovascular inflammation, neuroinflammation, and metabolic disorders like obesity and diabetes.
Solomon characterizes this evolution as a natural progression. "We started in immune-oncology, expanded to other oncology areas, then into immunology and inflammation, and now we’re moving into cardiovascular inflammation, neuroinflammation, and even obesity and diabetes," he explains. "The common thread is the immune system."
The "Plumbing" Problem: Why Big Pharma Struggles with Big Data
To understand why a 95,000-employee titan like AstraZeneca finds a startup’s services indispensable, one must look at the "plumbing" of modern drug development. Pharmaceutical companies are drowning in data, yet they often lack the infrastructure to translate raw patient samples into actionable clinical insights.
Solomon refers to himself and his team as "high-end plumbers," a metaphor for the complex, often messy, and expensive task of infrastructure-building that keeps the flow of innovation moving. "I fix very expensive plumbing issues," he says. These issues typically manifest as bottlenecks in the clinical trial process: poor patient stratification, inability to identify biomarkers for adverse events, or failure to optimize dosing schedules for complex combination therapies.
The Methodology: The "Immune MRI"
Immunai’s approach distinguishes itself from conventional AI in the life sciences sector. While many companies attempt to apply predictive algorithms to pre-existing, often noisy, legacy data, Immunai insists on generating its own "digital twins" from clinical patient samples.
The process is rigorous:
- Sample Acquisition: Biobanked patient samples from clinical trials are shipped to Immunai’s New York laboratory.
- Single-Cell Profiling: The samples undergo single-cell multi-omic profiling. This is the "Immune MRI."
- Data Digitization: For every patient, the team constructs a matrix of approximately 10,000 cells. Each cell is analyzed for 37,000 gene expressions, 75 surface proteins, and VDJ sequencing.
- Correlation Analysis: By profiling patients before and after therapeutic intervention, Immunai correlates these immune-system "snapshots" with clinical outcomes such as progression-free survival or overall toxicity.
Supporting Data: Resolution Over Scale
In the AI-pharma arms race, there is a common misconception that "more data" is always better. Solomon challenges this, arguing that without the depth of single-cell resolution, massive datasets are essentially "low-resolution photographs."
"A lot of big numbers in this field don’t actually lead to better decisions because the data was collected without depth," Solomon notes. "If you’re scaling a black-and-white photograph, you’ll never be able to see the difference between green and blue. If that’s the distinction you need to make, you’re stuck."
Immunai’s AMICA database currently holds over 300,000 samples, with 50,000 captured at the granular, single-cell level. This depth allows the firm to extract meaningful signals even from small clinical cohorts—sometimes as few as 20 patients. Because the foundation model has been trained on a massive, diverse base, every new, smaller cohort adds to the collective intelligence of the system, creating a compounding effect that enhances the accuracy of future predictions.
Official Responses and Cross-Functional Synergy
The operational reality of this collaboration is a massive logistical feat. With dozens of researchers on both the AstraZeneca and Immunai sides, the collaboration spans several internal departments.
"We work with multiple groups: people on the AI and data science side, people in translational medicine, and people in clinical development," says Solomon. "Each group covers different indications and therapeutic areas."
This cross-functional integration is essential for AstraZeneca’s massive portfolio, which currently manages over 100 Phase 3 clinical studies. By embedding Immunai’s platform into the clinical pipeline, AstraZeneca is not just outsourcing data analysis; it is effectively integrating an external "brain" into its drug development engine.
The industry at large is watching these developments closely. Recent moves, such as the April 2025 collaboration with the Parker Institute for Cancer Immunotherapy—which assembled one of the largest single-cell datasets for real-world immunotherapy research—and a separate multi-year partnership with Bristol Myers Squibb in early 2026, suggest that Immunai’s "plumbing" is becoming an industry standard for companies looking to de-risk their pipelines.
Implications for the Future of Drug Discovery
The implications of this partnership are profound for the future of precision medicine. If Immunai can reliably identify why a monotherapy fails, or predict which patients will suffer toxic side effects before a drug even enters a Phase 3 trial, the economic and human costs of drug development could be fundamentally reduced.
1. Shift from Trial-and-Error to Rational Design
By understanding the "immune profile" of a patient, drug developers can move away from broad-spectrum testing and toward targeted therapies that match a patient’s specific biological signature.
2. De-risking the Pipeline
With a $2.67 billion price tag on the average drug, any tool that improves the probability of success in late-stage trials is transformative. Immunai’s ability to provide a "digital twin" of the patient’s immune response provides a level of certainty that was previously impossible to achieve with traditional clinical markers.
3. Expansion of Indication
The success of this partnership in oncology and IBD proves the portability of the model. As the partnership moves into cardiovascular and metabolic diseases, it demonstrates that the immune system is the "common denominator" in human health, and that AI-driven immune profiling will likely become a prerequisite for success in all areas of drug discovery.
As Immunai continues to refine its "plumbing," the company is moving closer to a future where the immune system is not a mystery to be managed, but a predictable, digital map. For AstraZeneca, the repeated investment is a clear signal: the digital era of drug discovery has arrived, and those who ignore the plumbing risk being left behind in a sea of uninterpreted data.
About the Author
