Big experiments are only big if they can fail

Some reflections on Arena Bioworks' unexpected wind down as a fellow institutional experimentalist

When I saw the news that Arena Bioworks was shutting down two days ago, I was surprised and bummed. I generally don’t like to see teams fail, and I always feel an extra something for people who take a big swing for science.

But the thing that bummed me out more than anything was the way the science community responded. It was kind of gross to watch. The tone of the response over the last day or so has largely been glee. Social media snark dressed up as analysis and I told-you-so. As if someone else’s failure proved the old ways were right all along. There’s nothing more depressing to me than this kind of mean-spirited relief that the status quo is unchangeable.

Arena took a big shot. You don’t have to agree with their strategy, like them, or even trust their intentions to recognize that it takes some courage to exit a system you know to unlock something in unknown territory. It feels like we’ve become so used to scarcity and slow progress that anyone building differently is primarily treated as a threat, instead of a fellow human hoping for something better.

When efforts to break the status quo fail, we all lose a little. If we actually want to improve how science is organized, funded, and translated, we certainly can’t ridicule those trying. I want more people trying. Best case, they succeed. Worst case, there’s informative failure. I think those are pretty great outcomes for the larger enterprise of science, especially if someone wants to put their own private money into it.

We need more institutional experiments for science

I could see in some of the commentary an underlying false dichotomy that for-profit experiments in science compete with academic ones. Lots of “see this is why billionaires should butt out” vibes. I find this so confusing.

Don’t we want more resources for science? Don’t we want more avenues to enable basic research? I don’t understand why academic and private investments are mutually exclusive. I think experimenting with more for-profit investment into science, if done well, could grow the pie in several unique ways:

• Expand the kinds of science that can be done. Some basic research areas fall through the cracks of public funding. Market-aligned models can support ideas too applied for academia but too early for traditional biotech. Something in between purely basic and strictly applied science could potentially help with the “valley of death” challenge. This is not an easy problem, but it’s worth continued attempts to chip away at it. And anyways, science isn’t linear so who knows what differentiated fundamental insights might emerge at this intersection of basic and applied.

• Expand who gets to do it and how. For-profit places like Bell Labs (and Arena, Altos, Calico, Arcadia, etc) have drawn in engineers, entrepreneurs, and scientists who might never thrive in academia but have something vital to offer. And there are a wide variety of ways to structure these combinatorial teams and goals. We have grown accustomed to a monolithic path through university labs, but that can and should change. Science is big and heterogenous; our engines should be too.

• Expand sources of funding. Public and philanthropic dollars are essential but not enough. Private capital (when values-aligned) can fill critical gaps and help scale solutions and tools in a feed-forward way. It can also better connect basic research with real world utility, which isn’t a bad thing. It’s not the only thing, of course, but most people aren’t making that strawman.

It’s too bad that Arena’s experiment ended, in part because it felt like they never got it fully off the ground to complete the test. On the other hand, I have mad respect for poker players who fold early. Of course, we shouldn’t assume they are totally done yet. I know as much as everyone else on the outside, and STAT’s write-up suggests there are open paths they might pursue in the future. I genuinely wish them the best with that.

Arcadia’s own evolving model

If I had to give a generous interpretation to some of the glee, it might be due to external overconfidence. In recent years, new science institutions have kicked off with astronomical fundraising numbers and splashy PR, as if they’d already solved what generations couldn’t. The tone can feel hubristic and annoying in a way that invites scrutiny, even if some of that just stems from the pure excitement of it all.

As some of you know, I also co-founded a for-profit science institute called Arcadia Science. Building a new science institute that mixes research and business is no joke. Science is hard. People are hard. Business is hard. Bottom lines are real and vulnerable to market conditions. Together, they can be absolutely brutal. I haven’t woken up feeling like I’m good at my job in about five years.

When I read about Arena’s struggles with macroeconomic shifts that made their start-up studio model untenable, I felt a sting of recognition. We faced a similar thing at Arcadia. When we began designing our own commercial pitch back in 2019, the world looked very different. Capital was abundant, platform biotech was hot, and there was optimism that great foundational science could fuel a steady pipeline of spinouts. Compared to today, raising early stage funds was straightforward.

I’m guessing that we pursued a similar model to what Arena ultimately considered: leverage in-house science to seed start-ups, incubate them through early milestones, and reinvest later for value through equity. It looked elegant on paper. For a while, it even seemed feasible if we churned out enough shots per year. Then the markets turned.

When we opened our doors in 2021, raising capital was already becoming harder, milestone expectations higher, and investors wanted assets, not platforms. The idea of successfully spinning out several start-ups per year seemed ludicrous. It was clear by 2023 that the math wouldn’t work.

There was additional internal reckoning. The startup-studio model requires massive operational bandwidth: recruiting and training founders, managing teams across therapeutic areas, handling regulatory and fundraising complexity. We were new to all of this; merely getting to baseline performance took all of our energy. All while maintaining a coherent scientific program and talent.

To be honest, it just wasn’t a good model for us. I’m not sure that better macroeconomic conditions would have made a difference. In fact, I’m pretty sure the economic downturn saved us from a long, unproductive battle that we would have lost in the end.

This may have been specific to Arcadia, as the start-up approach didn’t play to our scientific strengths. We’re an evolutionary biology company doing deep R&D to understand how evolutionary patterns and relationships could be used to better structure and interpret biological data for drug discovery. It demands patience, iteration, validation, and real evidence of real utility. The path to revenue is certainly long and non-linear.

So we pivoted. It was relatively gradual at first. We initially shifted our math to one start-up a year to give ourselves more time to ice dead-ends and develop conviction, i.e. quality over quantity. It then slowed to one start-up every few years. And then finally, we called it. It didn’t make sense. We dropped the start-up studio concept and chose to focus almost entirely on our evolutionary platform. We’re still doing deeply fundamental science and you can actually read all about it here if you want.

These shifts were extremely difficult to confront. And even harder to execute. Lots of people were disappointed, the team shrank, certain flashier corners of the world stopped paying attention to us as the collective gaze shifted towards new institutions and AI buzz. I don’t think we thought we were failing, but it definitely didn’t feel like we were succeeding yet.

But honestly, it’s been great. It’s been liberating to ignore all the noise so that we can focus on the hard technical work at-hand. I’m more excited about our work than ever before. I’m also more confident that we have a shot at commercial success, although I’m even less clear on the details of our approach (in part because the world is changing so fast amidst the AI boom).

Arcadia by the numbers

Another aspect of the news about Arena’s wind down is how much they’ve spent. There’s not usually a lot of reporting about this for newer science institutions (or older ones for that matter). So I thought it might be worth sharing a bit about our burn as well. This section is mostly for others who are interested in institution-building and could use some help developing budgets.

Some notes:

• Arcadia has spent a little over $50M across the last five years for an average of about 40 people. That may sound high, but it’s roughly in line with a typical academic department or set of NIH R01 (directs + indirects) for a group that size. We also manage our own physical space and pay above market salaries.

• Our burn is a lot less than we originally anticipated, given that we raised $500M for 10 years. We imagined growing to about 100-150 people quickly. We abandoned that goal pretty immediately when we realized a smaller team allowed more flexible and exploratory work, especially in the beginning.

• I’m not exactly sure how much runway we have left in years, as we are actually in the middle of scaling some of our platform and exploring AI/ML workflows that will likely need more compute soon. That adds up quickly.

• Our peak size in 2023 was nearly 50 people, and we’ve since shrunk to 37, as of today. This is mostly due to the fact that we had to explore a lot of different directions over the first few years, which allowed us to identify a more specific set of things we had conviction about. This is not unique from most start-ups.

If there’s an overlapping lesson between us and Arena, it might be that being more frugal in the beginning is really critical for institutional learning and iteration. Of course, I am making a lot of assumptions here about Arena, so I can only speak for myself. Staying frugal early didn’t just stretch our runway; it made us nimbler, more self-aware, and better at killing bad ideas quickly. A long runway isn’t just financial security; it’s the space to experiment fully.

Despite this, almost all companies face a lot of pressure to grow and show evidence for revenue opportunities before you’re anywhere near ready. That shouldn’t be a surprise, and we didn’t go into this expecting otherwise. This is why we decided to keep our board and cap table extremely lean, so that we could ensure stakeholder alignment and withstand external pressures for the sake of our long-term science. But it hasn’t been easy, and I have a lot of empathy for other companies tackling hard, mission-oriented technical problems that have to grapple with this tension.

Lessons as the ultimate deliverables

Anyways, props to all the folks at Arena Bioworks. Ignore the haters. Arena’s struggles aren’t proof that institutional innovation is impossible; they’re proof of how difficult it is. Building new scientific institutions is science. Experiments, if done correctly, should have a very real possibility of failing. Otherwise, it’s not really an experiment.

In light of that, it would be awesome if Arena shared more of what they’ve learned from their experiment in the coming weeks, months, or years. In my mind, this is the ultimate payoff for any effort like this, regardless of whether you spend $5M or $500M. That’s the real privilege and opportunity that a swing like this gives you.

If Arcadia misses the mark one day, I hope I have the fortitude to follow through on what I’ve put forth here. It would be great if you could all celebrate that informative failure with me, encourage me to share more of our lessons at that moment.

So instead of piling on, I’m going to pour one out for Arena for shooting their shot on behalf of science.