Context: I work for one of the larger cultivated meat companies. I want to remain anonymous to avoid having to think about how what I’m saying might impact my company. There has been a lot of recent discussion about cultured meat feasibility, so I wanted to chime in with some mental models that I think are important for the bull case on cultured meat. This post is a bit crude—I thought it was more important to get the ideas out there than to spend a long time researching and refining.
Setting the stage
Many commentators have pointed out that a lot of apparent disagreement over whether cultured meat “will happen” flows from a misalignment on what “happening” means. In particular, what timescale, what market share, what revenue, etc.
Additionally, there are different types of players that care about different things. Here, I’ll focus on two in particular:
- The VC, who cares about making a return on their investment, within some payback time.
- The Altruist, who primarily cares about making sure cultured meat achieves a large market share rather than not, and secondarily wants it to happen sooner rather than later.
There are important considerations I think are missing from the EA conversation from the perspective of both actors.
Technological development and investment cycles move at different speeds
Transformative technologies generally take decades of substantial investment to develop. For example, solar energy took over 50 years and hundreds of billions of dollars of investment to reach where it is today: the cheapest source of new energy.
Investment cycles, on the other hand, are generally much shorter. They are based on the time horizons of the investors involved, as well as the natural boom and bust of hype cycles. The timing of hype cycles has a lot to do with human psychology, and how much disappointing progress an investment community can bear before the orthodox thinking becomes that the whole technology is doomed.
Technological development often requires multiple investment cycles from varied funding sources before it succeeds. For example, solar power went through multiple investment cycles from private and public sources before it became the success story that it is today. For example, oil companies like Exxon invested heavily in solar power technology in the 1980s, but lost faith within a decade. Later on in the 2000s, VCs invested heavily in Silicon Valley clean tech companies, but this investment bubble eventually burst with terrible ROI for most investors. In these cases, technological development moved slower than the investment cycle, leading to poor returns for the investors involved (Taming the Sun by Varun Sivaram has a great summary of the history of solar). However, The Altruist might see each of these periods of investment as important to the gradual cost reduction of solar power over time.
This makes “stepwise” development critical.
In order to maintain investor interest and avoid bubble bursting, it’s critical for companies to be able to demonstrate gradual progress and milestones. When a company has demonstrable progress, the investment cycle can extend to match up with the rate of technological development.
Part of demonstrating progress depends on the competency of the companies, but part of it also depends on the structure of the market itself. If there is a technology which is feasible in the long term, but there is no way to get there with successive demonstrations of progress, it is much more difficult to develop than one where you can demonstrate progress along the way.
One example is Genentech’s early somatostatin project. When recombinant protein technology was being developed, the most promising market was clearly insulin. However, insulin was a large, complex protein, that would have been infeasible given the nascency of recombinant technology. Instead, Genentech focused on somatostatin as their first project. Somatostatin had no market demand, but it was a small, simple protein that could serve as a proof of concept for recombinant protein expression. Once Genentech was successful with somatostatin, they were able to raise sufficient funds to eventually develop recombinant insulin.
It’s not a sure thing that a simple protein like somatostatin existed at all. Without such a protein that Genentech could use as a proof of concept, recombinant protein technology might never have been developed.
In the case of solar power, initial development was funded by a small market for photovoltaic panels on satellites during the space race era.
I think of an analogy of trying to cross a stream full of rocks–One parameter is how far you can jump, but another is how close together the rocks in the stream are.
Cultured meat's stepping stones
- Unlike solar energy, biofuels, or many other sorts of transformative technologies, cultured meat is consumer facing. This gives companies access to the powerful tool of marketing to make up for deficiencies in other areas. Strong marketing can cause consumers to pay a higher price for the same product. This is in contrast to an industry like energy, where the standards of success are more objective and decision are made by businesses with a bottom line. Effective Altruists might be prone to underestimating the potential effects of strong marketing.
- Consumers are diverse and heterogeneous, making it feasible to find customers at various stages of technological development. “Early adopters” are willing to pay a price premium and put up with product deficiencies because they are excited about the product.
- Cultured meat can start with hybrid plant-based / cultured products. Many companies are pursuing this strategy because the products can capture some of the taste benefits of cultured meat while keeping costs lower.
The cultured meat market’s ample access to stepping stones will make it easier to demonstrate progress and leverage this progress into successively larger investments.
For The VC in particular, this also means there are opportunities for substantial ROI that don’t involve price parity in the short term. A good analogy is Beyond Meat and Impossible Foods–neither are particularly close to price parity with conventional meat yet, but both have generated strong returns for their investors. In cultured meat's, there may be a successful company making hybrid products before cultured meat is near the cost of conventional meat.
Response to Rethink Priorities
Rethink Priorities recently published a meta-analysis of three techno-economic analyses (TEAs) on cultured meat feasibility. Their meta-analysis was thorough and well reasoned. This meta-analysis and associated research "has turned Linch from a cultured meat optimist to being broadly pessimistic. Neil wants to be more agnostic until further research from Rethink Priorities and others.”
In particular, Humbird’s analysis presents a forceful case against the feasibility of cultured meat. However, in my opinion, his analysis presents only a moderate reason for worry for the VC, and a smaller reason for worry for The Altruist.
For the VC, cost parity is not a necessary criteria for generating substantial ROI (cf Impossible Foods and Beyond Meat).
For the Altruist, achieving substantial market share doesn’t necessarily need to happen within the current VC investment cycle (although success would be better sooner rather than later).
As the authors note, all three TEAs have flaws in their methodology. This is because there is no rock solid methodology for arguing against the ultimate feasibility of a transformative technology. Especially for the Altruist, who can stomach longer time scales, not finding an easy path right now is not a solid reason to think that there won’t be one in the longer term, when there’s billion of dollars of R&D funding, and the global supply chain has years to shift around the incentives of the new market.
An alternate way of assessing the feasibility of cultured meat is to imagine a path of stepwise development from now to a world where there are enough milestones that tens of billions of dollars have been invested. At that level of investment, there are sufficient resources to break fundamental assumptions that would be in any TEA made today.
Taken this more holistic view on the future development of cultured meat presents a more hopeful picture than simply looking at current TEAs.
Hi avacyn,
Thanks for the engagement. I'm honored that you devote a section to your post just discussing our summary. Here are some rushed thoughts (Note that I'm speaking for myself and not Neil or Rethink Priorities, I think there's a ~50% chance that Neil substantively disagrees with me on at least one important point here, and a >90% chance that someone on the RP team does):
This is not, strictly speaking, accurate. The Altruist cares that altruistic goals are met, which cultured meat having a "large market share" may or may not be in the path of. When referring to cultured meat within an EA context this usually but not always means reduced degree of animal suffering in factory farms[1]. But there may be other ways of achieving this:
The question for altruists isn't "cultured meat or not", it's (as usual) "is marginal investment in cultured meat the best use of resources to accomplish our altruistic goals"
Re: your theory of change, this sounds plausible (at least conditional upon cultured meat being the correct path forwards).
One thing I'm unsure of is whether it makes sense for altruistic resources on cultured meat to be spent now (instead of invested to be spent later). I wish I had more conceptual or empirical clarity about how much e.g. early solar cell investment was really in the causal path to having solar cells today (as opposed to e/g. a technological overhang from other advances making later development of solar cells a lot easier), and how much we think this can reasonably generalize into future technologies. I think figuring this out is important for not just cultured meat but other technological advances EAs might care about as well.
Note that Humbird's target price and market of $25/kg at 100kilotonnes/year (so a market of $2.5B/year) was indexed on the higher end of plant-based meats, not conventional meat.
I don't think this is a fair paraphrase of our analysis. I think we make clear that the magnitude of the errors in CE Delft seem larger, and they all point in the same direction (suggesting motivated reasoning).
I will note that the current front page of the CE Delft report (archive) describes it thus: "... explored various avenues for cutting costs when production takes place in full-scale plants, realized in 2030.
We conclude that substantial cost reductions that bring CM production costs close to the benchmark are feasible. This requires a combination of reductions that covers nearly all aspects of the business case." [emphasis mine]
One might naively conclude from their report and the resulting publicity that the report was talking less about the ultimate feasibility and referring to a more specific and nearby date.
I think this is actually a pretty important crux. You take it as a given that those resources will be enough, but The Counter article that precipitated our summaries put it thus:
I think it's a pretty important crux here: in worlds where EA literally devote almost all our (present and future) non-longtermist resources into cultured meat R&D, which in the more optimistic scenarios may mean tens of billions of dollars and cajoling multiple Nobel Prize-level scientists, is this enough resources and technical know-how to get us to truly cost-competitive and widescale cultured meat in the next e.g. 20 years? (at least barring political and consumer acceptance concerns)
I think if you believe the path forwards is 0-3 Nobel Prizes worth of breakthroughs plus some moderately hard (say Tesla-level) engineering/logistical challenges away, this will imply a very different worldview and forecasts than from someone that thinks "work worthy of many Nobel prizes" is a non-hyperbolic summary of their current understanding of the situation.
[1] alternative reasons include climate change, feeding the poor.
Protein from seaweed (which GFI is now interested in), methane consuming single cell protein, hydrogen consuming single cell protein, leaf protein concentrate, and maybe even electrically powered single cell protein, though it looks like they are better at producing vinegar than protein.
Thanks for helping to extend the list!
Thanks Linch! Apologies for the things I misunderstood / misrepresented about your report. Any sloppiness was the result of being rushed. I hope it’s clear that I was trying to engage in good faith :).
I agree with your characterization of the Altruist’s goals. Indeed, I think one of the biggest reasons to be bearish on cultured is if you’re super bullish on plant-based!
Re my theory of change, one other area that we may take lessons from is AI. I’m hesitant to speak too much about that since folks here know way more about it than me, but if AGI ends up being possible, it seems like it would have been important that domain-specific AIs were highly economically useful at various levels of sophistication.
Re the crux you identified - I agree that imagining such scenarios is important to thinking about the possible outcomes of cultured meat. I’m not sure I understand your distinction in the last paragraph of your comment. However, a couple of related thoughts: