MIT Method Yields mRNA Vaccines That Survive a Year at Room Temperature – So Far, Only in Animal Studies
Researchers at MIT have used an AI method called AGENT to develop mRNA vaccine formulations that withstand two months at 37 °C and up to a year at room temperature. The results are promising – but so far demonstrated only in animals – and the cost savings are estimates, not proven gains.
The News
An MIT research team has published a study in Nature Biotechnology, covered on September 28, 2026, describing a new method for making mRNA vaccines thermostable – that is, able to survive at room temperature without deep freezing. According to the study as reported by Inside Precision Medicine, the method produced solid-form formulations of mRNA and lipid nanoparticles (LNP) that retained full bioactivity after more than two months at 37 °C, and remained stable at room temperature for up to a year (79e8af7b).
The senior authors are Ana Jaklenec and Robert Langer, who is the David H. Koch Institute Professor at MIT. The lead authors are PhD student Jinbi Tian and postdoc Khanh Tran. The work was carried out in collaboration with MIT CSAIL (fcdf5948).
How AGENT Works
The method is called AGENT – Algorithm-Guided Experimental design for lipid Nanoparticle Thermostabilization. It combines high-throughput experiments with Bayesian optimization, an AI technique well suited to learning from relatively small datasets. The researchers started with nearly 50 FDA-approved excipients, and the algorithm used the experimental results iteratively to predict increasingly promising formulations. The optimization was completed in six iterations over one month, which according to the researchers substantially reduces trial and error (79e8af7b).
Jaklenec emphasizes that small datasets are precisely the point: "The real beauty of this algorithm is that we can use it with small datasets. It's very hard to run thousands of experiments, so this algorithm makes it easier for us to achieve formulations with the properties we want – in this case, stability" (fcdf5948).
It is worth noting what "AI" means here: not a large language model, but a statistical optimization method that selects which experiments to run next. It is a narrow, but highly practical, application.
What Has Actually Been Shown – and What Has Not
The most important caveats are preclinical and regulatory. All the results come from animal studies, not human trials, and no regulatory authority has approved changes to current vaccine storage requirements (79ddb79e). The researchers also observed structural changes, so-called "blebs," in the reconstituted LNPs, and point out that standardized methods are needed to determine how such changes affect product quality and performance. Manufacturability and regulatory questions require further research (79e8af7b).
There is also no timeline for when – or whether – the technology could reach clinical use.
Another uncertainty concerns current storage requirements: the sources give different temperature ranges for conventional mRNA vaccine storage, and it is unclear how precise current requirements actually are.
The Animal Studies and the Relevant Formulations
The method was tested with two clinically relevant LNP compositions, representative of those used in Moderna's and Pfizer-BioNTech's covid-19 vaccines. In studies in rodents and non-human primates, the thermostable formulations produced immune responses that were not inferior to current solutions, including delivery via microneedle patches in primates (79e8af7b).
Reuters reports that mice vaccinated with the thermostable particles after long-term storage showed immune responses equivalent to mice given vaccines in lipid nanoparticles resembling Moderna's original formulation (6b3d7d92).
The Economic Estimates – and Why They Are Uncertain
The study authors cite earlier estimates that thermostable vaccines could reduce storage costs by 71 to 86 percent and cut waste by at least 50 percent (79e8af7b). These are not measurements from this study, but references to existing estimates – and they reflect why the cold chain is a real problem: deep freezing requires costly infrastructure, and breaks in the chain can ruin vaccines.
What the Researchers Believe the Technology Could Be Used For
Tian sees a broader application than mRNA vaccines alone: "Our approach expands the application not only for mRNA vaccines, but also for therapeutics or advanced drug delivery platforms such as controlled-release particles or microneedle patches, which require the formulation to be either in solid form or stable at higher temperatures" (6b3d7d92).
What Remains
The study is a concrete example of AI-guided formulation search compressing months of trial and error into a single month – but it is still preclinical research. The next steps would be clinical trials, large-scale manufacturability, and regulatory approval. None of these is in place. Until then, the most precise statement is that the MIT team has demonstrated the principle, not the solution.