Stéphane Bancel: It's the first time there is a cancer vaccine working. The field has been doing that for 20 plus years, more than a thousand clinical trials that have all failed.

Jorge Conde: What was different this time? What is it about mRNA technology that enables the immune system to learn in a way that other approaches were unable?

Stéphane Bancel: We all have cancer cells all the time in our body. Our immune system is very well trained. Basically notice those cancer cells very early and get rid of them. But if your cancer grows, then the question is how can you retrain immune system? We're going to basically take a biopsy of your tumor. We're going to read all the letters of its DNA. And then we're going to do the same things with a healthy cell of your body. And we're going to literally compare letter by letter, nucleotide by nucleotide. And then we're going to use an algorithm to identify which one of those mutations are the most relevant. And so when this is injected in your body, it teaches the immune system the signature of your cancer cell that it missed.

Jorge Conde: What gets regulated here? Because every dose is different. So obviously every dose doesn't get approved.

Jorge Conde: Hi. Um, welcome to the a16z podcast. Uh, I'm Jorge Conde, a general partner on the a16z Bio + Health team. I am thrilled today to welcome back Moderna CEO Stéphane Bancel. Um, for folks that have been longtime listeners may recall, uh, Stéphane joined us on the a16z podcast back in December of 2020. Um when we were talking about all of the work Moderna did to uh bring us the the mRNA COVID vaccine and at the time um if you can go back and listen to that episode you'll hear um how quickly um Moderna was able to react to the existence of the virus to analyze it uh to essentially print a vaccine um that would protect people against against the the COVID-19 uh vaccine and hence put us in a much better position uh with respect to the to the COVID pandemic. We titled that episode "The Machine that Made the Vaccine", and the reason why we're talking again today in August 2026 uh is because Moderna has uh put out news uh on a really big advancement on what they can do with mRNA technology when it comes to cancer. And so I want to hand it over to you uh Stéphane. Again a very big welcome. Thank you for coming back. But maybe the place to start is let's uh let's lead with the news. Um you uh Moderna and Merck uh announced uh earlier this month in August that uh you had uh conducted a phase three clinical trial in melanoma and have gotten uh very encouraging results. So why don't I hand it over to you and tell us uh what have you announced? What have you seen in this phase three trial and then I do want to dig in into what Moderna is doing in cancer.

Stéphane Bancel: Wonderful. So Jorge, thank you so much for having us back. We're very happy to be to be with you. So indeed uh last week now uh we shared the news with our colleagues at Merck that after working 10 years on a individualized uh treatment against cancer using mRNA technology that the phase three was positive. It's a it's a first of many. It's the first time that there is an agent in melanoma that is better than Keytruda alone so it's a big deal for patient of course with melanoma. It's the first time there is a cancer vaccine working. Uh the field as you know has been doing that for 20 plus years. I think more than a thousand clinical trials that have all failed. Um and um if you look at the phase two data because what we announced last week is that we met the primary endpoint of the study which was recurrence-free survival i.e. people having the cancer back or dying that we met.

And then to our own surprise because this was the first interim analysis—this is not the end of a study, it's a first interim of a study—and what was a surprise even to us is we met the secondary endpoint which was distant metastasis-free survival which is of course takes more time to mature because it means that you have distance metastasis from your primary tumor and we also met that uh endpoint which was great again unexpected from our side but again means it is really good. Um we will share the data very soon at a big medical oncology conference which is as you know what the field does. But just to give you a sense maybe to orient kind of directionally we showed at ASCO the big you know oncology conference in the spring of 2026 a few months ago that our phase two study which was also randomized against Keytruda alone—so the same type of study—showed that around 50% of people had recurrence-free survival versus people only getting Keytruda and this is 5 years after treatment. And as you know in oncology five years is considered by doctor like a cure and so it's a big big deal. And if you look at the data from overall that phase two you had around 80% of people that were disease-free five years after their treatment and their surgery of melanoma. And so we're very excited about what it means for the field. Uh we are working very hard already with regulators to file so that the drug can be available to patient as soon as possible and we hope in 2027 and we're going to try to make it as fast as we can. Uh the factory is ready in Massachusetts where we have figured out how to make and scale a product for every human being at a time.

Jorge Conde: Well, there's I mean there's so much to to unpack here. So this is an extraordinary uh advance uh for the the treatment of in this case melanoma and hopefully over time cancers more broadly. Uh so let's unpack some of this if we could Stéphane. Uh the first one um uh talk to us uh for folks that are less familiar with cancer treatments. Uh why is Keytruda what is Keytruda and why is Keytruda alone not sufficient? Why was Moderna necess—or the the this particular mRNA cancer vaccine necessary in conjunction with Keytruda?

Stéphane Bancel: Sure. So if you look at Keytruda which is kind of one of the leading immunotherapy that people might have heard of is basically if I oversimplify for non-biologist it's basically a molecule that basically open the gates for letting the dogs out that are going to go and attack your cancer from your immune system if you want. Um the thing about checkpoints is when they work they're fantastic because those people 5 years after treatment are cured in the sense they have no remaining disease. Uh but only 60% of people are disease free after five years if you look at the phase three published data of Keytruda. So again for those 60% of people it's amazing but it means that there's 40% of people where you go through the treatment you're fighting for cancer and the treatment doesn't really work for you and also what is difficult is those treatments are wonderful but they a lot of time come with very serious side effects very different from chemotherapy or radiotherapy. The side effect of immunotherapy are mostly immune disease.

So you see people if you look at the label or the clinical studies people that get checkpoints whether it's anti-PD-1 or anti-CTLA-4 that have a lot of T-cells expanded. But if you take my analogy, you unleash the dogs in a dark room. And they're going crazy, they are barking, they are running around, and they attack whatever because they don't know what they are looking for. And in many cases they attack the self which create those autoimmune disease and they don't attack the cancer.

Jorge Conde: Mhm.

Stéphane Bancel: So that we could design a product to basically teach if I go back to the dog analogy to teach those dogs what to look for very specifically. And that's really the beauty about our technology is if you think about Keytruda basically unleash the dogs but our mRNA individualized cancer treatment gives the picture of what the dog should look for and says: "This is what you're looking for, go get it and destroy it."

Jorge Conde: Okay. Okay. So now that that's where the personalized cancer vaccine um uh technology will make a big impact on the treatment of melanoma. So that we'll come back to the personalized piece because I think that's just fascinating not only from a technology and business perspective but biology perspective. But before we get there, help me define what is a vaccine here. Because as folks think about vaccines historically, they think about it in a preventative context—you get a shot to prevent polio or prevent COVID. Here you have cancer, so it's a therapeutic vaccine. How should people think about that distinction?

Stéphane Bancel: It's a great characterization and the reason the field has used the word vaccine—it was not us using it, we just follow the field as I mentioned there been a thousand plus clinical trial—is because it's about teaching your immune system. If you think about a COVID vaccine or flu vaccine, you teach your immune system so it's ready when the virus comes to you.

Here, whether it's outside factors or just as you have cell replication and you have mistakes that happen that create mutation of DNA that are cancer cells, our immune system is very well trained to basically notice those cancer cells very early and get rid of them. But if your cancer grows, then the question is how can you retrain the immune system? So I think that's why the field use the word vaccine for that approach as well. Even as you said it's a therapeutic treatment approach post cancer, it's about the teaching of the immune system.

Jorge Conde: Okay. And so as you point out, the field has tried this many times before. A thousand on the order of a thousand clinical trials um have have tested this theory. They've all failed. Uh you and Moderna and Merck have succeeded here. Um what was different this time? What is it about mRNA technology that enables the immune system to learn in a way that other approaches uh were unable uh uh to be successful?

Stéphane Bancel: Yeah, I think there's there are two components. Uh, one is the mRNA technology and I think the other one is the individualization by designing a product for one human at a time. So, let me go through those two.

On the technology of mRNA side, what we have known and published uh actually with Karolinska as you know the institute in Sweden that give the Nobel Prize of medicine back in 2015 is that with our technology—I cannot speak about other mRNA companies that have different mRNAs, different lipids and so on—but with our technology when we inject our mRNA in a muscle whether it's a COVID shot or flu shot or this cancer treatment basically the mRNA goes down to your lymph node and enters the APCs, the antigen presenting cells, which as you know are key component of your immune cells, and the mRNA gets inside the APCs. This we've demonstrated and proven at the time with Karolinska. And then it basically translates the message contained in the mRNA inside the APCs, inside your immune cells, and presented from within. So we think it's a very important differentiation from most of the previous vaccine in cancer in the field that were made by protein or peptide that basically are made in reactors, are injected in the in the in the patient, but basically turn into the blood. Because as you know recombinant or protein when you inject them they just go into your blood and they turn around, so your immune system sees them but not in the same manner from within as it done with the mRNA. So we think that's one very important component of uh immune presentation if that makes sense.

The other component is really individualization. In the past a lot of time people have tried with non-mRNA i.e. protein technology or peptide, but also we tried a shared antigen. Whereas here what we said is because cancer is a disease of DNA...

Jorge Conde: Mhm.

Stéphane Bancel: What we're going to do here because the cost of sequencing have dropped so much in the last 20 years is we're going to basically take a biopsy of your tumor. We're going to read all the letters of its DNA, the 3 gigabytes of of these genes. And then we're going to do the same things with the healthy cell of your body. And we're going to literally compare letter by letter, nucleotide by nucleotide. And then we're going to use an algorithm to identify over your hundreds or thousands of mutation which one based on the current knowledge of a field of immunology and cancer which one of those mutations are the most relevant. We select the 34 that we believe are the most relevant and we stitch them together into one big mRNA molecule which we make in 30ish days for you that is injected then in a hospital intramuscularly. And so when this is injected in your body, basically it teaches the immune system the signature of your cancer cell that it missed. Not the signature of every other patient with a shared antigen, but the very specific signature of your cancer cell.

And what we we showed at ASCO and we published from a phase two study, but we believe it's the same thing in the phase three because it's mechanistic, is that around 90% of the antigen are different patient to patient. Because when we started we had no idea because again the field came from shared antigen. So when we started like we have no idea we're going to get 2%, 5%, 90% of same antigen across all the patient. Actually 90% of antigen are different from a human to another one.

Jorge Conde: Wow. So the only way this can work is through personalization, individualization.

Stéphane Bancel: Exactly.

Jorge Conde: You know assuming that that carries over. Um, and so in that regard, uh, you know, if you're doing, let's say, this normal to tumor comparison of of of the genome, you find the differences. I'm curious how you arrive at 34 as up to 34 as the right number. Um, I'm sure there's a very good reason for that. Um, but what is the algorithm that enables you to do that? Is this something that's proprietary to Moderna? Is this something that is known within the field? Help us understand like help us look into that black box.

Stéphane Bancel: Sure. So I want to share a little bit about the black box—not too much because there's a lot of know-how and and things are very confidential to us—but basically we started with of course what is known in the field. And so we basically use a lot of database and publication and a lot of scientists and doctors kind of best-in-class in immunology and in oncology. And then from that starting point we used a lot of internal data that we generated over time. We also partnered with some companies that because they are having the diagnostics space or they are let's say cell therapy and other space in oncology had access to a lot of data, a lot of T-cell mapping and so on that were very useful for the learning.

The thing that is interesting about the data we showed last week is this is what I consider mRNA-4157 version 1.0. Because the algorithm from the phase three was the same from the phase two, was the same from phase one. But because we've been doing this for 10 years, it's a 10-year-old algorithm. So as you look at the data it works pretty well, right? As we said about the phase two data 80% of people are disease-free after 5 years, it's amazing for those patients, but there's still 20% of patients that don't respond. And so one of the thing we're going to be doing now that we have access to a phase three patient data and samples is to go back and mine that data to figure out why some patient responded and why some other patient did not respond. Because we have access to all their blood sample, the sequence, everything, and we're going to try to see can we improve the algorithm. Uh and we will go to the FDA if we find scientific reason why we should change the algorithm to go from let's say 1.0 to a 2.0 algorithm and then change it. Of course, we have to do that in a very controlled way to ensure we don't lose efficacy very obviously.

But the way I think about it is like when we talk about AI, we always joke that the current version of AI is the worst we're going to see in our lifetime. Well, it's exactly the same for this mRNA therapy, which is the current version of mRNA therapy at Moderna is the worst version you're going to see for the rest of medical history. And so that gives me a lot of hope not only in melanoma for those 20% of patient that don't respond, but also for potentially other tumors that have been really hard in the field like you know pancreas cancer and others where immunotherapy doesn't work. We want to be able to learn a lot about the technology using also what the field has learned in the last 10 years. Because the field has learned a lot as you know, this is even not in 1.0. So that's why I'm so excited about what's coming next.

Jorge Conde: That's fantastic. So, um, looking back, uh, you and I have had, we've known each other for a very long time. Uh, I won't depress you or me by saying how long,

Stéphane Bancel: But you were very young. You were in kindergarten! [laughter]

Jorge Conde: But so I've had the benefit of seeing Moderna from the earliest days. Yes. And one thing that is true that was true then is true today is first of all you are an engineer at heart and you have from the very beginning been obsessed with process, with operations, with being efficient. And those pieces are really essential here. Um, and let me maybe one way to frame it is I think a lot of people think about the other big personalized therapy that exists in cancer is CAR-T cell therapy, right? And in that case, CAR-T cell therapy for folks that may be less familiar, you're taking cells from a human, you're genetically modifying them, you're growing them up, and then you're putting those cells back into a human. And there's a lot of failure points, there's a lot of complexity, it's very expensive, and it takes a long time. Here you're doing something very different. Talk to us about the operational feat that is personalized cancer vaccine manufacturing. What does that needle-to-needle time look like? How do you think about that and how do you execute on that?

Stéphane Bancel: Sure. So the time is around 42 days now needle to needle. So from taking the biopsy to getting the vaccine in hospital ready for you. I think we're going to be able to improve that as we still have a lot of efficiencies to work on and automations and so on. We are aiming for 30-ish days, and then I think over time we'll be able to get to a couple of weeks.

If you think about the steps: you have the biopsy, you ship the sample to us, we do the sequencing of the tumor and the sequencing of the healthy cell from your blood, then we do the run the algorithm, then we need to do the DNA of the 34 mutations, then you make the mRNA from that DNA template, then you formulate it with a lipid nanoparticle (LNP), then you put it in a vial, you test it for release, and you ship it back to the hospital.

And the big difference with CAR-T, which was your question, is we have no cells in our process. It is a cell-free process. Where in CAR-T you have cells and big reactors and you have big volumes because as you know the reason you have big volumes in biotech industry is if you compress the cells too much they die. You have the same issue with CAR-T, so it's all everything is huge. In our case, because we don't have cells, the volumes are tiny. The concentrations are very high. And so the equipment is very small, which allows us to automate a lot of the process using robotics and software.

Jorge Conde: And at scale, how many roughly how many for let's say let's just focus on melanoma how many doses would you need to produce in a given year?

Stéphane Bancel: So we've already got thousands of doses because of nine clinical studies that are ongoing. The facility will be able to make—the one in Marlborough, Mass—tens of thousands of doses. And then as we keep improving the technology that number is going to go up in the same facility and then we might need to build several other facilities. But if you look at the incidence of melanoma, if you have tens of thousands of doses you're going to cover easily the melanoma market.

Jorge Conde: Yeah, I can believe that. Have you disclosed how you think about COGS and price or is that something that...

Stéphane Bancel: We have not disclosed yet. We need first to disclose the data with our Merck colleagues. We need to engage with the payers once we can share the data in terms of what is the value being driven there and so on. But this has not been discussed yet.

Jorge Conde: So maybe one place to focus is on this concept of personalization. I'm sure you've seen the story of the GitLab founder who went founder mode on his own osteosarcoma. Number one: do the future Sids of the world come to Moderna, or is this N-of-1 phenomenon something that you think will just happen and exist in parallel?

Stéphane Bancel: So I think they will come most of them to Moderna because it is just going to be easier and safer. Because as you know making an injectable product always carries risk of contamination of a product. If you inject to somebody a product that has even one copy of bacteria, you might give the patient sepsis. Then you always have a question of quality because when you have a multi-step process, a mistake can happen. And of course, if it's industrialized and has been validated in terms of Good Manufacturing Practice kind of FDA standard, you have much less chance of this happening.

So, it's a bit like every tool in life, which is: do you make your first knife because there's no knife store and you are in a cave and you need to feed your family? Yes, of course you make your first knife because you have to feed your family, right? But when you have a store making high-quality knives down the street, you're going to use your time to do something else. So I think it's a bit of the same phenomenon which is like in any technology: when you have industrial scale of high-quality product, you use your time as a human to do something else with your time.

Jorge Conde: And in that world, how does the regulatory apparatus function here? In other words, you mentioned earlier you along with Merck will prepare a regulatory filing soon. What gets regulated here? Because every dose is different, so obviously every dose doesn't get approved. Is it the process for synthesizing mRNA? Is it the algorithm? Is it a combination of the entire system? Help us understand that and build intuition around how we think that these kinds of personalized medicines will be regulated in the future.

Stéphane Bancel: Yes. So the good news is there are precedents as you mentioned CAR-T. CAR-T was also approved in the same way as we believe this mRNA will be which is as a process BLA not a product BLA. So as you know Moderna has five products approved, so that's really product approval. On this one, the whole process since we started in the clinic the IND was a process IND.

Because we had to ask the FDA can we go to the clinic, do you think it's safe and do you think we have a good control of a process so we can do safely a phase one study. So we already had that discussion just to go into the clinic years ago. And then before we started every phase three we need to have end of phase two meeting and agree the design of a study, the manufacturing protocol with the FDA. So those discussions have happened for years. There's been a lot of discussion, a lot of engagement. Several times we've had technical questions on the manufacturing front where we basically requested additional meetings to ask the guidance, to also educate them on the technology, what we've learned and so on. So there's already been a lot of discussions and there's a very clear regulatory pathway in terms of approving the entire process.

Basically, what the FDA wants to know, which is very legitimate, which I would want for my own family sake obviously, which is: if you get the same sample at the beginning—the tumor and the blood—do you get the same product made at the end of a big black box? And that's what we have to demonstrate first to ourselves and then with the data to the FDA so that we have really robustness overall process. So if we have the same input, we're going to get the same output going to patient as an individualized medicine.

Jorge Conde: How do we think about moving beyond or how do you all think about moving beyond melanoma? Is this an approach that's going to be applicable to a broad range of cancers? Are there cancers that are much more likely where this is going to be a viable option versus others? And sort of what's your—I'll use the word—what's the ambition here for where cancer vaccines can happen?

Stéphane Bancel: The ambition is pretty big because we believe we have demonstrated at least to ourselves and we hope to the world that we are able to create an education of T-cells. And this we showed it even this year: we took the blood before treatment, after treatment of cancer patient melanoma with our technology, and we showed that not only we have an expansion of the T-cells, but we have de novo new T-cells being created that recognize what we code in the mRNA that was not in the patient's body before the treatment. So we really in my book have proven to ourselves and to the clinical community that Moderna's technology can teach the immune system to develop new T-cells to go attack your cancer.

So based on that there are basically I would say three different vectors we're going after in terms of expansion from melanoma:

First, we're going everywhere where Keytruda works. Because as I told you we believe the mechanism of action of a PD-1 and Moderna's therapy are totally orthogonal. So we think these allow to have synergistic element and efficacy for the patients. And so we are in phase three for lung. We're in phase two for kidney cancer, bladder cancer. So we have a whole slew of studies ongoing where the world knows that Keytruda works, and we believe you're going to see a material improvement versus Keytruda alone.

The second vector is to go early in disease where checkpoints work. And the best example is we announced in the spring of 2026 starting a phase three study for patients with stage one lung cancer as a monotherapy without checkpoint.

And we are doing that because we believe when you go early in disease checkpoints are not used because of the side effect that they bring. Because if you have stage one cancer, the medical field thinks it's worth monitoring your cancer versus giving you a checkpoint because not everybody's going to respond, but everybody's going to get pretty serious lifelong side effects like autoimmune disease. But what if you could have an mRNA made for cancer patient that has lung disease stage one, which you can find easily with screening? The idea is you do a surgery which is standard of care, and you give our monotherapy which the side effect is similar to a vaccine—you know you might feel tired for a day but that's it.

The third vector is where checkpoints don't work, so of course it's where you have the highest risk. But because the mechanism of action is different from a checkpoint, we and Merck believe that there's a very good scientific rationale to go try. So two places we're trying right now is pancreas cancer and also gastric cancer. Those two cancer types checkpoints and Keytruda do not work. But we think because again the mechanism of action is different from checkpoint and we know now that we have a proof that we can create de novo T-cells, we think it's an experiment worth running. And as you know literally yesterday Revolution Medicines had a wonderful new medicine approved for pancreas cancer using the KRAS mutation. What if you could combine that medicine and our therapy? Those are very orthogonal mechanisms of action.

Jorge Conde: Wow. Well that that gives a lot of reason for I think cancer patients and their families to have a lot of hope for the future of novel therapies in this field.

Stéphane Bancel: Yes, and on top of what we just said remember this is version 1.0. So what I saw is very powerful and I'm really pushing our team to think really outside the box and to do a lot of analysis and to use AI to look at that gigantic set of data that we have, which is what are the things we can learn from the clinical studies to understand about the people that did not respond. Because I think you always learn more from things that don't work than things that work. So I want to obsess about the 20% of patients that do not respond 5 years out so we can understand why they did not respond and can we tweak anything in the algorithm or in the technology to be able to help them.

Jorge Conde: Wow, that's remarkable. And just to wrap, I think it's remarkable to see how you were able to take a technology platform that originally wasn't built for a pandemic, point it at a pandemic, create a vaccine for millions and millions and millions of people, and essentially point it back towards treating some of the diseases that you had originally intended 10 years later as you described.

Stéphane Bancel: It's very remarkable. And the piece that's going to be exciting or maybe to close is before the end of the year we should have our pivotal study—so late stage study—for rare genetic disease for kids that have rare genetic disease of the liver. The phase 1/2 has shown kids three years on drug doing fantastic.

And in June we had our annual science day and we announced that the next mountain where we are pointing mRNA platform is autoimmune disease. Because if you think about it we've learned a lot from infectious disease which are mediated by the immune system, and cancer we just been talking a lot about immune system. So we learned so much about the immune system that we think we have some very novel approach on how to treat the root cause of autoimmune disease not the symptoms which is what the pharma industry has been doing. It's of course very helpful to patients to treat the symptoms so we can have higher quality of life, but it doesn't treat the root cause. And we think we might have found ways to use the immune system to treat the root cause of autoimmune disease. So we're quite excited about what's to come.

Jorge Conde: Would the theory there be that you'd have personalized autoimmune modulators or would this be more a product or more a process?

Stéphane Bancel: So, we're doing both. What we presented in the spring was a product that would be the same for everybody. But what I'm the most excited about which is in the lab still is the ability to do individualized autoimmune treatment where you target directly to the immune cells that are attacking your body as self when you have an autoimmune disease and to have basically part of your immune system going attacking those immune cells that are out of order so that you are able to take the cause of immune disease out. Again, it's still early days but that's what I'm excited about today.

Jorge Conde: Well, going from infectious disease to cancer to eventually autoimmune disease, we would love to have you back on the podcast to film episode 3 and complete the trilogy when you're ready.

Stéphane Bancel: Wonderful.

Jorge Conde: Stéphane, thank you so much for joining us on the a16z podcast. As always, it's great to see you, and congratulations!