00:00Why are you doubling down and what is the money being used for?
00:04Well, first of all, thanks so much for having us here.
00:06It's a really exciting time for the company.
00:08We're coming out of this year with a series of research breakthroughs
00:12where we've now shown that we can use AI to design molecules
00:15that we never even thought was going to be possible.
00:18So we're really using this money to double down on that research agenda.
00:21The models show no signs of slowing down.
00:24And then also really growing the team and starting to take these models
00:27and start deploying them really at scale.
00:29And starting to create new molecules that, again,
00:31we never thought would have been possible to create.
00:35Elena, it feels like this is a relatively pivotal moment in AI's push forward
00:41on drug discovery and on healthcare more broadly.
00:44You're someone who invests in life sciences.
00:47What is it that made CHI stand out that you thought it needed follow-on money
00:51just three months after a Series A?
00:53Absolutely.
00:54So CHI stood out because 2025 was the year of AI discovery.
00:59And what CHI allows you to do is understand and predict the interaction between molecules
01:06to design and develop new medicines.
01:09And the reason we invested now is because we believe 2026 will be the year of deployment.
01:15And pharmaceutical companies that adopt this technology will absolutely leapfrog those who do not.
01:21And it's a moment where medicines that could not be discovered any other way will now be discoverable.
01:29And the aha moment, in many ways, was following the release of CHI 2, Josh.
01:34Just tell us why that was so important, the latest antibody generation model that you released
01:39and I'm sure suddenly had a load of inbound calls on the backup.
01:42Yeah, so of course this is our Series B.
01:45But if you look back at our Series A, we told our investors that over the next two years or so,
01:50we would try to reach a 1% success rate on this antibody design task,
01:54meaning that 1% of the molecules coming out of remodels would function in the lab.
01:58We thought that would be a breakthrough moment in the field.
02:00But a couple months ago with CHI 2, we actually announced that we could do this with about a 15% to 20% success rate.
02:07So I think really leapfrogging again what even we thought was achievable in the past year or so.
02:13So that's really been, I think, a big moment for the field.
02:15And now moving us away from just research curiosities into models that are really useful for actually discovering drugs.
02:24Elena, that's sort of mind-blowing for many who aren't in the weeds on life sciences as much as you are,
02:30that there is only a 1% hit rate was going to be really tangible for the field, and let alone 15% to 20%.
02:38Talk to us about what have been the limitations thus far in drug discovery, in labs, for pharmaceuticals,
02:44and what this change is in terms of AI.
02:47Absolutely.
02:48So when researchers try to make a new model, a new molecule,
02:52it's essentially searching for a needle in the haystack.
02:55It happens in a lab.
02:56There's a lot of variability.
02:58And it's really challenging.
02:59That's why even though we've seen breakthrough medicines and many people benefit from them,
03:04there are still lots of areas where people get sick and have no options.
03:09What CHI has done is show that they can design computationally medicines that have over an 85% success rate
03:19in achieving basic drug-like properties.
03:22There's additional research to do, work to be done, that takes a drug from the computer into the lab,
03:27into people, into clinical trials.
03:30But there's been a longstanding hype around AI for drug discovery.
03:35And now when we meet pharmaceutical executives and they meet CHI, everyone says, this is the year of deployment.
03:43Now these models work in a way that they will actually contribute to new medicines that make it to patients.
03:48And, of course, Josh, you've previously been backed by Thrive Capital, Menlo Ventures,
03:53OpenAI has been a strategic investor and a place you used to work.
03:56But with General Catalyst and with Oak on site, I'm sure those meetings with pharmaceuticals come more thick and fast.
04:04But what more is the money being put towards when you talk about scaling this project?
04:08How much compute do you need and how much you need to invest in GPUs as well as the engineers
04:14and the fierce talent wars to get the right people in the door to build the models too?
04:19Yeah, well, I'm thrilled to be working with General Catalyst and Oak here.
04:22You know, this round is actually about a lot more than the money.
04:26You know, you bring together some of the top investors in this space in health care.
04:30If you look at the folks joining our board, Annie Lamont and Hemant Naja,
04:34these are investors who have been Midas-less health care investors, really plugged into the pharma space.
04:39We're here with Elena as well, who's just such an incredible thought partner on really building out the business.
04:43So the capital is going to work among all the things that you mentioned.
04:46We're really putting an astronomical amount of these costs into compute.
04:51We're trying to turn compute into better molecules.
04:54We run a ton of lab experiments to really test whether the models work in the real world.
04:58And we're doing a lot as well in order to start deploying these models,
05:02building the right kind of software tooling in order to build this computer-aided design suite for molecules.
05:08There's the talent wars as well.
05:09I'm very, very grateful to be working with some of the top folks in the industry here.
05:13When we started the company, we actually brought together a number of heads of AI from top AI drug discovery companies.
05:18And we've only continued to build out an exceptional team here.
05:21We're a small but mighty team, but we're really putting this to work in order to turn science from a biology from a scientific discipline
05:30into one that looks a little bit more akin to engineering.
05:33Yeah. Elena, you said a great line that basically there's been a lot of hype around AI and how it will be applied to health care.
05:41But what do you think helps cut through to those who are laymen, to a lot of the real details and data being shown in this,
05:48that we will see productivity gains, we will see real deployment in 2026.
05:54What's your hope as to what this looks like next year?
05:57Well, I think for next year, the hope, and I think pretty strongly the reality,
06:03is that pharmaceutical companies will be able to discover medicines more efficiently.
06:07So Josh mentioned there's a lot of compute going in.
06:11That's really on the research side.
06:13What's incredible is this technology is efficient.
06:17It is helping people discover medicines.
06:19And to me, the most important thing that laypeople will see is when we use these tools,
06:26we can discover medicines that take advantage of known biology and impossible chemistry.
06:31You could not make the molecule.
06:33You could not make the medicine without using this technology.
06:37And what people will see are medicines that are curative that would not have been discovered any other way.
06:43And that's where laypeople will really see the benefit is in medicines that have impact on their lives.
Comments