Are AI-discovered vaccines the next big frontier in pharma?

Why did MSD’s deal with Evaxion turn attention toward AI-discovered vaccines rather than a single licensing story?

Merck & Co., Inc. (NYSE: MRK), through its global arm MSD, made headlines by exercising an option to license Evaxion Biotech A/S’s (NASDAQ: EVAX) AI-designed vaccine candidate EVX-B3. The structure was familiar: a $7.5 million upfront payment and up to $592 million in milestones, plus royalties if the product ever reaches market. Yet the underlying signal was novel. EVX-B3 was not discovered through conventional immunology or pathogen surveillance but through Evaxion’s AI-Immunology platform, which models immune interactions and predicts epitopes that traditional methods might miss.

The deal extended Evaxion’s cash runway into 2027 and gave MSD responsibility for development costs. Investors saw it as a validation that computational immunology has matured into a credible sourcing channel for pharma pipelines. The fact that MSD also extended its evaluation period for a second Evaxion candidate, EVX-B2 for gonorrhea, underscores that this is not a one-off experiment but a systematic test of AI’s ability to deliver viable vaccine assets.

What scientific advances in immunology and protein modeling are making AI-discovered vaccines credible in 2025?

For years, vaccine discovery relied on a combination of epidemiology, trial-and-error, and incremental improvements. The COVID-19 pandemic accelerated the adoption of mRNA platforms, but the bottleneck remained antigen choice. Today, advances in AI-powered epitope prediction are changing that equation.

Deep learning models, including transformers and graph neural networks, are now benchmarking significantly above earlier heuristic-based systems in predicting B-cell and T-cell epitopes. Reverse vaccinology 2.0 uses genome-wide pathogen data, filters with AI scoring systems, and proposes candidate epitopes with higher accuracy, reducing wasted cycles.

Simultaneously, structural biology has been transformed. AlphaFold 3, which can predict not only proteins but also complexes involving ligands and antibodies, allows researchers to simulate how vaccine antigens may be presented to the immune system. While not sufficient alone, these advances give developers a powerful triage step before investing in animal studies, improving both speed and probability of success.

Which companies are defining this frontier, and how are pharma giants aligning their strategies?

Evaxion is only one node in an emerging ecosystem of AI vaccine discovery. BioNTech SE, already known for its COVID-19 mRNA vaccine, deepened its AI capabilities by acquiring InstaDeep. This move integrated AI not just for target discovery but across manufacturing, logistics, and trial design, underscoring a holistic commitment to AI as a platform.

Baseimmune, a UK-based startup, is developing what it calls “future-proof vaccines.” By modeling likely pathogen mutations, it designs synthetic antigens intended to remain effective even as viruses evolve. MSD’s venture arm co-led Baseimmune’s Series A round, giving it early exposure to the company’s antigen-design approach.

DIOSynVax, a University of Cambridge spin-out, takes a different tack. Its work is focused on “broadly protective” antigens, computationally designed to provide immunity across entire lineages, such as betacoronaviruses or avian influenza. Supported by funding from CEPI, it reflects a public-private push to create vaccines that anticipate rather than chase pandemics.

Together, these companies illustrate the diversity of AI-driven approaches—from epitope prediction to variant anticipation to lineage-level protection. The common denominator is that each is now attracting serious pharma and public health backing.

What bottlenecks could slow the transition from in-silico vaccine design to clinical reality?

The enthusiasm for AI-discovered vaccines is tempered by several realities. The first is biological translation. Even the most accurate algorithm cannot guarantee that a predicted epitope will produce a strong, safe, and durable immune response in humans. Animal models and early clinical data remain essential.

The second is manufacturability. Not every promising antigen can be expressed, folded, or stabilized at scale. Adjuvant selection and vaccine format—whether mRNA, protein, or vector—also affect outcomes.

The third bottleneck is regulatory acceptance. Regulators will require the same safety and efficacy evidence as with traditional vaccines, and AI’s role will be seen as supportive, not substitutive. While agencies may welcome AI-augmented rationale, they will continue to demand conventional validation before approval.

These realities mean AI shifts where risks and costs are concentrated—it accelerates early discovery and triage—but it does not eliminate the long and expensive clinical development path.

How are investors reading the milestone-heavy financing structures common in AI vaccine deals?

The MSD–Evaxion structure reflects a cautious but increasingly common template in biotech. Upfront cash is modest, protecting the larger company, while milestone payments provide upside if technical and regulatory hurdles are cleared. For Evaxion, the $7.5 million upfront is meaningful: it extends operations without diluting shareholders, while the nearly $600 million in contingent payments represent transformative upside.

Investors responded by pushing Evaxion’s shares up more than 25 percent on announcement. Yet sentiment remains speculative. Institutional ownership in Evaxion is low, and its stock has fallen over 70 percent in the past year. Analysts give it a “Strong Buy” rating with a $11 target price compared to current levels near $4, but volatility and retail-driven ownership mean the stock can swing sharply on data or deal news.

The broader lesson for markets is that pharma prefers options and staged capital deployment. Startups benefit because they gain runway without raising equity, while retaining the chance to cash in on success.

What does a realistic roadmap for AI-discovered vaccines look like over the next three years?

The most likely breakthroughs will occur in three areas. First, pathogen-specific vaccines for antibiotic-resistant bacteria or elusive viral targets are poised to move into early human trials, offering direct proof of AI-guided discovery. Evaxion’s eventual disclosure of EVX-B3’s target and its IND submission will be a bellwether.

Second, computationally designed “variant-proof” antigens for influenza and coronaviruses will continue to gain public and philanthropic support. DIOSynVax’s entry into Phase 1 with its universal flu concepts will test whether AI can deliver breadth beyond incremental updates.

Third, the fusion of epitope prediction with structure-guided design will allow increasingly sophisticated preclinical design, reducing attrition rates. As AlphaFold-era models expand and integrate with generative approaches, the cycle between AI predictions and wet-lab validation will tighten, creating adaptive learning loops.

Rather than one dramatic inflection point, expect a steady shift in vaccine R&D pipelines as AI integration becomes standard.

How do Evaxion, BioNTech, Baseimmune, and DIOSynVax compare in approach and strategic positioning?

Evaxion’s strength lies in its epitope discovery engine, with programs across infectious diseases and oncology. BioNTech’s approach is enterprise-wide, embedding AI into every operational layer. Baseimmune emphasizes anticipatory design against pathogen evolution, and DIOSynVax focuses on lineage-level broad protection.

Pharma behavior is consistent across these partnerships: commit modest upfront capital, secure optionality on promising antigens, and link milestone payouts to de-risked progress. Governments and coalitions, for their part, use grants and public funding to accelerate computationally designed vaccines that serve global health priorities.

A key differentiator will be modality flexibility. Platforms that can take AI-discovered antigens and encode them across mRNA, protein, or vector formats will attract the most attention from pharma partners and procurement agencies.

What should readers and investors watch to separate real signal from hype in the AI vaccine narrative?

The next phase of validation will come from disclosures and early clinical results. Evaxion must eventually reveal EVX-B3’s target and show data from preclinical and early human studies. DIOSynVax must prove its universal concepts deliver breadth. Baseimmune will need to move its computational antigens into tangible programs with pharma. BioNTech will need to demonstrate that its AI integration translates into accelerated programs beyond COVID-19.

The sector is past the “hype” phase. AI is no longer a buzzword in vaccine discovery—it is embedded in deal structures, R&D roadmaps, and funding strategies. The winners will be those who combine computational power with translational discipline, ensuring that predicted epitopes become safe, manufacturable, and effective products.

For investors, the frontier of AI-discovered vaccines represents both risk and opportunity. Companies like Evaxion offer high volatility and potential upside tied to validation. For pharma, option-based deals provide low-risk exposure to emerging platforms. For public health, the prospect of variant-resistant or broadly protective vaccines could redefine preparedness.

The question is not whether AI-discovered vaccines are the next frontier in pharma—they already are. The question now is which platforms will clear the gates of clinical and regulatory proof to turn AI-born designs into the next generation of commercial blockbusters.