9 Companies Using Artificial Intelligence to Fight Infectious Diseases
Development of a novel drug is a tedious process that requires immense investments of resources and time—on average, it takes from 8-10 years and capitalized R&D costs are reported between ~$1.1B median and ~$2.2B+ per asset in large-pharma cohorts for a new medicine to complete the journey from initial discovery to the marketplace. And then patents give pharmaceutical companies the right to be the sole supplier of a new drug in a certain country to make such time and efforts investments justified. But if we are talking about a rapidly developing new plague, time resources at the disposal of the pharmaceutical industry appear to be scarce and new solutions to emerging challenges have to be found faster.
For example, during the pandemics remdesivir and dexamethasone were shown to have efficiency in fighting COVID-19, but it led to supply shortages for both medicines. To avoid the collapse of drug supply chains, scientists at the University of Michigan, Ann Arbor, Timothy Cernak, and colleagues turned to Artificial Intelligence (AI). They used a commercial drug synthesis AI program Synthia to find the most efficient and cost-effective strategy for synthesizing medicines. Researchers programmed Synthia to look for new synthetic solutions for 12 medications that at the time were being tested as potential COVID-19 therapies. In 11 out of 12 cases Synthia was able to find novel synthesis solutions, and for one of the studied compounds, the software came up with 4 different synthetic strategies.
What about other pathogens menacing the existence of the human race? In 2019, there were an estimated 4.95M deaths associated with bacterial antimicrobial resistance, including 1.27M deaths directly attributable to resistance. And as the response to this threat, only 17 new systemic antibiotics have been approved by the FDA since 2010. Existing methods for screening new antimicrobials are very costly, require a significant time investment, and are usually limited to a narrow spectrum of chemical diversity.
Scientists at MIT decided that Artificial Intelligence could give an answer to this issue. With the help of machine learning algorithms they have identified a powerful new antibiotic compound halicin. In laboratory tests, the drug killed many of the world’s most problematic disease-causing bacteria, including a WHO-priority pathogen A. baumannii, a bacterium resistant to all known antibiotics that has infected many U.S. soldiers stationed in Iraq and Afghanistan. In 2023–2024, a separate deep-learning campaign yielded a new structural class with activity against MRSA and VRE and efficacy in mouse models, demonstrating how model-guided triage can surface non-obvious chemotypes.
Overall, AI in infectious disease is shifting from proofs-of-concept to reproducible pipelines—identifying antibacterial and antiviral leads, informing synthesis and manufacturability, and feeding public-health systems with earlier signals. The sections below update the companies applying these approaches today.
AbCellera
Vancouver-based AbCellera (Nasdaq:ABCL) is a public company building an AI-driven antibody discovery stack that combines high-throughput microfluidics, single-cell sequencing, machine learning, and computational design. The platform is used across infectious-disease programs, including pandemic response and parasite transmission-blocking antibodies.
Computational models are integrated with experimental selection and optimization, while peer-reviewed outputs illustrate the platform’s rapid antibody discovery capabilities in virology and parasitology. Beyond viruses, AbCellera co-authored work identifying human monoclonal antibodies that block malaria parasite transmission—an application of its high-throughput, ML-guided antibody discovery to antimicrobial targets.
In 2018, AbCellera was one of four groups named to the pandemic rapid response project at the Defense Advanced Research Projects Agency (DARPA). The project was aiming to build a platform that could create a field-ready antiviral within 60 days of isolating a virus.
With Eli Lilly, the company discovered COVID-19 antibodies—including bamlanivimab and bebtelovimab—that progressed to EUA and large U.S. procurement before EUAs were later revoked as variants shifted. It also partnered with NIAID’s Vaccine Research Center and Ichor Medical Systems to couple broad-neutralizing antibody discovery with nucleic-acid delivery for rapid outbreak response. Beyond that, AbCellera and AbbVie have expanded their work together to develop new T-cell engagers for cancer, linking AbCellera’s discovery platform with AbbVie’s oncology programs.
Anima Biotech
New Jersey-based Anima Biotech applies its Lightning.AI platform to infectious disease by coupling high-throughput live-cell imaging with neural networks that learn mRNA-biology features and triage small molecules that block viral protein synthesis.
The PathwayLight toolset (TranslationLight, TranscriptLight) “visualizes” pathway-level events like splicing, localization, and decay to link phenotypes with mechanisms. In its RSV program, Anima screened ~100k compounds and processed ~20M images in the cloud to flag series that reduce RSV protein translation and viral load without blocking global translation.
The company frames its lab+ML runtime as a “Biology GPU”—an execution layer for AI models to “compute” inside cells—reportedly used across >20 discovery programs via pharma partnerships.
Partnerships extend this AI stack into large portfolios. In 2018, Eli Lilly signed a multi-target translation-inhibitor deal for $14M research funding. Takeda partnered in 2021 to discover mRNA-translation modulators for genetically defined CNS diseases. AbbVie joined in 2023, paying $42M upfront with up to $540M in milestones across three targets, leveraging Lightning.AI for oncology and immunology.
Atomwise
Established in Toronto in 2012 and later moved to San Francisco, Atomwise applies deep learning to structure-based small-molecule discovery via its AtomNet platform—screening at scale against protein pockets to prioritize novel chemotypes for medicinal chemistry. This technology reportedly excels at understanding complex concepts as a combination of small pieces of information, which is differentiated from other techniques on the market by its ability to find and optimize novel chemical matter.
Atomwise applies its AtomNet deep-learning models to “try on” vast libraries of small molecules in 3D protein pockets and rank which ones are most likely to bind—an AI alternative to traditional high-throughput screening that a 2024 multi-target study reports as viable across 318 targets.
During the COVID-19 pandemic, Atomwise partnered with global teams during COVID-19 to pursue broad-spectrum candidates using AtomNet.
In 2020, Atomwise established a joint venture with RNA-virus researcher Nito Panganiban (Tulane) to discover small-molecule inhibitors against flaviviruses (Zika, dengue, West Nile) and a broader RNA-virus superfamily using AI-guided screening and chemistry. Tulane materials note ongoing work on a dengue protease inhibitor program with Atomwise. Atomwise is also extending its workflow through pharma partnerships, including a five-target discovery pact with Sanofi with $20M upfront and milestones.
A2A Pharmaceuticals
New York-based A2A Pharmaceuticals was founded in 2016, and since then the company has been designing computationally optimized small molecule therapeutics for the treatment of cancer and antibiotic resistant bacterial infections.
The company applies its SCULPT platform—a fragment-based, computational design workflow that builds target-specific virtual libraries at “thousands to millions” scale with iterative optimization and ADMET filtering—to generate small-molecule leads, using multiple computational resources and AI. This shifts early discovery from modifying known chemotypes to producing candidates computationally tailored for complex targets from the outset.
The company also has previously established a joint venture with Insilico Medicine with the creation of Consortium.AI (2018) to discover small molecules—originally announced for DMD/rare diseases.
Their antibiotic programs target biosynthetic enzymes unique to Gram-negative bacteria to avoid human off-targets and reduce resistance risk.
Partnerships and recognition include JLABS program participation describing an iterative combination of docking and supervised machine learning to optimize libraries, plus collaborations the company lists with groups former leadership from Pharmacyclics, Novartis, J&J, and P&G.
In antiviral medicine, A2A entered a collaboration with Laxai Life Sciences to design SARS-CoV-2 main-protease (Mpro) inhibitors using A2A’s SCULPT/AI workflow.
Evaxion Biotech
Copenhagen-based Evaxion Biotech (Nasdaq:EVAX) builds vaccines with an AI-Immunology platform that ranks protective antigens from whole pathogen proteomes and integrates design modules used from in silico selection to preclinical testing; the same stack also powers its oncology programs.
Evaxion’s AI-Immunology platform ranks vaccine targets directly from whole pathogen proteomes, using EDEN for bacteria and RAVEN for viruses to predict antigens that can drive protective B- and T-cell responses; top-ranked candidates are then taken into wet-lab validation. The company has updated EDEN with a toxin-antigen predictor to better prioritize bacterial vaccine components and has added an automated vaccine-design module intended to speed translation from target lists to construct designs.
Recently, Evaxion introduced an automated design module for its AI-Immunology platform that replaces manual vaccine-construction steps, enabling faster, more consistent antigen-sequence and structural optimization for both new and existing vaccines.
In infectious disease, the company out-licensed EVX-B3 to MSD and kept an option structure around EVX-B2 for gonorrhea, both discovered with this AI stack, while also partnering with Afrigen to explore an mRNA format for EDEN-selected gonorrhea antigens. In Staphylococcus aureus, EVX-B1 showed protection in a large non-rodent surgical-site infection model, while in cytomegalovirus, EVX-V1 combined AI-discovered antigens with an AI-optimized pre-fusion gB that improved neutralization and reduced infection in preclinical models.
Insilico Medicine
Hong Kong–founded Insilico Medicine applies its end-to-end Pharma.AI stack—PandaOmics for target discovery and Chemistry42 for generative small-molecule design—to infectious diseases alongside fibrosis and oncology
Insilico reports 22 preclinical candidate nominations from 2021–2024 with an average 13-month path to Preclinical Candidate. In May 2022 Insilico announced the nomination of a novel preclinical therapeutic candidate for treating COVID-19, designed using their generative chemistry AI platform Chemistry42. The new drug candidate is a 3CL protease inhibitor unique from existing drugs in its class because it can be rapidly produced in sufficient industrial amounts upon need. One of Insilico Medicine’s AI-designed candidates for idiopathic pulmonary fibrosis (IPF) recently advanced through Phase 2a in the US/China.
Over the same period, Insilico also licensed out assets, for example Menarini’s Stemline licensed an AI-designed preclinical cancer drug from Insilico for solid tumors with $20M upfront. Additionally, recently Insilico raised $110M Series E and unveiled Supervisor—a bipedal humanoid lab robot that learns by imitation to handle routine wet-lab tasks (pipetting, reagent handling, instrument operation) and feed results into its AI discovery workflow.
Phare Bio
Cambridge, Massachusetts–based Phare Bio, founded 2020, is a nonprofit “social venture” building a generative-AI platform for novel antibiotics, developed with the Collins Lab at MIT and Harvard’s Wyss Institute. The group focuses on priority Gram-negative pathogens and moves AI-designed hits through preclinical de-risking before partnering or spinning out for clinical development.
Phare’s AIBiotics engine is described as a generative-AI discovery stack that designs new chemotypes, screens orders-of-magnitude more molecules at lower cost, and then layers ADMET/tox filters as candidates advance from hit→lead→preclinical. The ARPA-H-funded TARGET project is expanding this with millions of new training datapoints, up to 10 new model “filters,” and a future open-source database for AI-antibiotic discovery. This year, Phare Bio also secured added support from Google.org’s Gen-AI accelerator.
Recently, Phare Bio advanced two AI-designed antibiotic candidates—NG1 for drug-resistant gonorrhea and DN1 for both gonorrhea and MRSA—into preclinical development following laboratory confirmation of their activity. The Cell paper and accompanying GitHub repo detail the generative framework and the 24 synthesized molecules that yielded seven actives, including NG1 and DN1.
PostEra
Boston and San Francisco-based PostEra builds AI tools for medicinal chemistry: Proton, its generative chemistry platform that closes the design-make-test loop, and Manifold for retrosynthesis/search across purchasable chemistry with Manifold exposed via API and integrated into third-party discovery suites, such as the 2023 Post Era and Optibrium collaboration.
Proton is the core of PostEra’s multi-year Pfizer AI Lab collaboration, expanded in Jan 2025 to $610M and a new ADC track where Proton is applied to optimize cytotoxic payload properties. The company also has a five-program small-molecule deal with Amgen.
The same toolchain underpins PostEra’s antiviral work: Post Era co-founded and leads the open-science COVID Moonshot, applying ML design and retrosynthesis to SARS-CoV-2 Mpro inhibitors that advanced to preclinical development, supported by an NIH AViDD award of $68M.
PostEra’s pipeline page shows candidates beyond anti-infectives: four wholly owned programs in women’s health—two in polycystic ovary syndrome plus separate reproductive endocrinology and fertility assets—are in discovery, while the partnered portfolio lists four preclinical programs across obesity and oncology.
Peptilogics
Peptilogics is a clinical-stage biotech based in Pittsburgh, Pennsylvania, developing engineered antibacterial peptides. Peptilogics’ Nautilus platform couples generative models that propose peptide sequences in novel chemical space with predictive models that rank them by multiparameter drug properties for in silico triage.
The stack runs on a custom company-owned high-performance computing cluster system with a Cerebras partnership that reportedly cuts training time by up to 1000×, then feeds prioritized designs to a custom peptide synthesizer to generate ADME, pharmacology, PK and safety datasets for iterative retraining. The framework is generalizable to linear or cyclic peptides, including nonstandard amino acids, and it layers development context—indication, manufacturability, IP—on top of sequence/structure data to prioritize targets, including clinically validated yet difficult or previously undruggable classes.
Peptilogics’ pipeline centers on zaloganan (PLG0206), a locally applied antimicrobial peptide for prosthetic joint infection (PJI) that holds QIDP, Orphan Drug, and Fast Track designations, with a pivotal Phase 2/3 RETAIN DAIR trial planned to enroll 240 knee-PJI patients starting December 2025. A recent $78M Series B2 financing supports the pivotal zaloganan study.
Additionally, the AMR Industry Alliance profiles Peptilogics’ engineered cationic antibiotic peptide (eCAP) approach as a route to novel-class antibiotics with activity against Gram-negative pathogens. Beyond PJI, the company lists an orthopedic expansion of PLG0206 and PLG0301 for cystic-fibrosis–related infection/inflammation in preclinical development.
Topic: AI in Bio
