Summary

The core upshot is that even if you grant a high probability that each of these bottlenecks could be solved, the probability of them all being solved in coordination could be low^[1]. Moreover, the bottlenecks AGI seems least likely to resolve alone—regulation, politics, and consumer trust—could be those that take the longest to address.

Finally, if you agree with the perspectives we describe here, advocacy that helps animals in the near-term may not be irrelevant shortly before and after AGI. If we want highly competitive alternative proteins to be ready when AGI-level capabilities arrive, those bottlenecks need attention now.

Intro

Many think that our best bet for drastically reducing the number of animals on factory farms requires great alternative proteins that consumers will routinely eat instead of meat. Those who think there is a distinct possibility that transformative AI or AGI arrives soon may also think that AGI will solve cultivated meat and get it to market. 

The intuition is understandable. If AI can do world-class science, engineering, and strategic reasoning, surely it can crack the remaining problems in cultivated meat? But this treats cultivated meat as primarily a science problem. In practice, getting competitive alt proteins to market requires solving a chain of bottlenecks—many of which are institutional, political, and social rather than technical. 

Building on other writing on this topic, here we walk through everything you'd need to believe to think that AGI solves cultivated meat. This matters for guiding efforts now to set us up for the best chance of success post-AGI. If something won’t be solved under AGI due to specific bottlenecks that already exist, perhaps those bottlenecks need more focus now.

For AGI to solve cultivated meat, all of the following must hold

Cultivated meat (or animal welfare) is prioritized

Even if AGI can solve cultivated meat problems, its efforts have to be pointed there to ensure it does. This is a resource allocation question, and it seems uncertain regardless of how autonomous the AGI is.

How much this matters depends on how accessible AGI is. In a world of abundant compute where AGI-level tools are widely available, prioritisation is less of a constraint. Millions of agents could work on millions of problems simultaneously. In that world, alt protein companies, researchers, and advocates could direct AGI at cultivated meat challenges themselves, without needing to compete for access. In this scenario, the prioritisation concern largely dissolves.

But there are some plausible worlds where AGI capability is more concentrated—controlled by a small number of governments, corporations, or platforms. In those worlds, cultivated meat competes for AGI attention with cancer, climate change, poverty, national security, and profit-making (like making factory farming even more efficient). If today’s patterns hold, the sector has modest near-term commercial returns, and active political opposition in multiple jurisdictions. It is unclear if cultivated meat would rank highly on the priority list of those directing AGI.

Under more autonomous or less human-constrained AGI, the question is whether the system would choose to prioritize cultivated meat itself. There are multiple reasons an AGI might find cultivated meat to be beneficial, including food security, climate change, and animal welfare. Whether it would act on those reasons depends on how it weighs competing interventions, how heavily it weighs non-human welfare, and whether it models cultivated meat as the most effective lever among many. 

There is also a question about whether autonomous AGI systems would choose to prioritize animal welfare. There are reasons they might—food security, climate, and animal suffering all point toward reducing factory farming. But whether an autonomous system would weigh non-human welfare heavily depends on how it reasons about morality, and whether it inherits current human values or reasons from first principles. However, animal welfare is not well-represented in AI alignment frameworks, and the variation in views of animals across models (which could influence the views of future AGIs) could suggest there is not a particularly strong basis for assuming an AGI would prioritize helping animals.

Overall, in some post-AGI worlds, cultivated meat gets attention without any special effort. In others, advocates may need to actively steer AGI resources toward alt proteins. It's worth being aware that the case for advocacy may not disappear under AGI.

The remaining science is solved

There are several key open scientific problems in producing cultivated meat. A non-exhaustive list includes: Optimising animal-free cell culture media (previously fetal bovine serum was widely used), producing whole-cut meat analogues via scaffolding,^[2] and making cell lines and bioprocesses perform consistently at industrial-scale volumes.  While these remain constraints, they are increasingly well-characterized and progress is happening.

A few years ago, key questions remained open: whether animal cells could proliferate reliably in animal-free growth media, whether the resulting product could match the taste and texture of conventional meat, and whether any of this could work outside tightly controlled lab conditions. Speakers at a recent sustainable protein conference noted that most of these basic feasibility questions now have affirmative answers. The challenge has shifted to scaling—producing cultivated meat not in grams, but in tonnes. The science side of scale-up^[3] also seems like an AGI-tractable problem.

Current AI tools in the biosciences can already accelerate parts of this work—predicting scale-up dynamics, reducing wet-lab trial-and-error, and shortening validation cycles. The main constraint for today’s AI tools is training data: companies guard cell line data, media formulations, and process parameters as intellectual property, and the large-scale datasets that would power breakthrough models largely do not exist yet (though some work exists to make progress here).

But AGI-level systems could plausibly generate their own data through cycles of hypothesis, experiment, and refinement, or be so capable of generalisation that limited domain-specific data is not a hindrance. If the bottleneck in cultivated meat development were purely scientific, the case for "just waiting for AGI" could be strong.

Manufacturing scales affordably

Even if AGI solves the underlying science, cultivated meat must still be manufactured at scale. Producing enough to meaningfully displace conventional meat requires physical infrastructure—bioreactors, purification systems, supply chains—that is not yet widely available. One analysis estimated that producing around 0.3% of projected 2030 global meat production would require about 11 to 44 times the current bioreactor capacity of the pharmaceutical industry, at current levels of cell-culture productivity. AGI-driven advances in bioprocess engineering could substantially improve cell density, media efficiency, and bioreactor throughput, lessening this gap. But even a several-fold improvement in productivity could still leave the industry requiring a massive physical build-out from today’s very low baseline.

AGI could help here by designing more efficient production facilities, optimizing supply chains, and, through advances in materials science, reducing the cost of some inputs. 

But even with perfect designs, the physical work of building out facilities could still take some time. How long depends in part on how quickly AGI transforms physical industries. Some expect AGI-directed robotics and construction to dramatically compress build-out timelines. Others may think that real-world frictions will keep physical building slow. Either way, there is at least an initial lag between the arrival of AGI and any physical transformation it enables. 

Infrastructure also costs money. Better science and better-optimized facility designs could derisk investment in cultivated meat, which could attract more private capital. And if AGI drives broad economic growth, both philanthropic and private funders would have more resources available. Whether that money flows toward cultivated meat depends on the priorities of those funders—and, possibly, the AI systems advising them.

On balance, AGI likely improves the economics of cultivated meat manufacturing and could accelerate physical build-out, but it cannot change the fact that the industry is starting from almost nothing.

Regulators speed up, maybe by adopting AI

Regulatory approval is one of the largest bottlenecks cultivated meat faces today. At a panel at the Bezos Centre for Sustainable Protein Conference 2026, panellists were given a hypothetical choice between fast-track regulatory approval, a 100,000L bioreactor, or a massive drop in media cost. Most chose regulatory approval—because it unlocks revenue, investor confidence, consumer feedback, and it's the thing most outside of their control. Several cultivated meat leaders report similar views.

The crux is not whether companies can produce better dossiers. AGI could help substantially on that side by digesting regulatory guidance, ensuring compliance, conducting literature reviews, and designing and analysing toxicology and safety studies. Some companies are likely already using frontier LLMs for operational tasks like these (here's some evidence). Better dossiers reduce back-and-forth with regulators, and an AGI system could model approval pathways across jurisdictions, identifying which safety data satisfy multiple processes at once. 

But the primary cause of long wait times sits on the regulator side. The question is then whether regulatory bodies themselves would adopt AGI into their review processes. There are some early signs that some might.

The US FDA is reportedly deploying AI tools across the agency and said it completed its first AI-assisted scientific review in mid-2025. The UK FSA contracted AI firm Aiimi under a two-year contract to help staff process new product applications and identify food safety risks, and reportedly uses generative AI to clean and categorize data from food alert systems. While it's unclear if any of these examples apply directly to the novel foods regulatory process that cultivated meat would face, they suggest some willingness to use AI tools within regulatory bodies. 

Even so, some constraints limit how much this matters. First, food safety determinations almost certainly require a human decision-maker—both as a matter of legal accountability and because agencies carry significant reputational risk if an AI-cleared product later causes harm (e.g. see here and here). Second, if AGI dramatically accelerates the science and manufacturing of cultivated meat, the number of products seeking approval could increase substantially, potentially overwhelming existing review capacity even if each individual review is faster.^[4] Finally, whether existing regulatory approval frameworks would apply to cultivated meat produced via AGI-optimized methods is unclear. 

A more pessimistic view is that processes would only be allowed to be sped up so far because regulators have strong incentives to move cautiously on novel foods. Even if AGI produced a technically flawless safety dossier, agencies may still impose somewhat long review timelines to manage reputational risk.

The net result is that AGI can accelerate background tasks—literature synthesis, data processing, cross-jurisdictional mapping—but the human decision-maker at the end of the review process remains. And because approval is jurisdiction-by-jurisdiction, the overall picture depends on the slowest-moving markets. For cultivated meat to reach a scale that meaningfully affects farmed animal welfare, it needs approval across many jurisdictions. Even in an optimistic scenario where a handful of regulators move faster, that constraint remains.

Political opposition reverses

Multiple US states (Alabama, Florida, Mississippi, Montana, Nebraska, and others) and EU countries (Italy, Hungary) have banned or restricted cultivated meat production and sale. Several more have proposed bans. Here’s what the landscape in the US currently looks like:

The EU recently banned the use of multiple meat-related terms for the sale of plant-based proteins and cultivated meat (while ‘burger,’ ‘mince,’ ‘sausage,’ and ‘nuggets’ are allowed, terms like ‘chicken,’ ‘beef,’ ‘pork,’ and ‘bacon’ are not). 

As Lewis Bollard notes:

Let's say that AGI solves cultivated meat for us. Cultivated meat is already illegal in seven^[5] US states. It might soon be illegal in the entire European Union. By the time we get AGI, will they even be able to sell it anywhere?

These bans are primarily driven by agricultural lobby pressure. There is no obvious mechanism by which AGI reverses these political dynamics directly. If anything, if cultivated meat becomes more viable and widely produced, you could just as reasonably expect greater pushback from the agricultural lobby.

That said, if AGI makes cultivated meat commercially successful in jurisdictions where it remains legal, the opportunity cost of maintaining bans rises. States or countries that permit cultivated meat could attract industry investment, jobs, and tax revenue that non-permitting jurisdictions forgo, and food distributors operating across borders may push back against a patchwork of restrictions.

How fast this economic pressure translates into political change is unclear, and current trajectories in the US and EU point toward more restrictions before the likely arrival of AGI. Each year of delay until bans are overturned is a year in which conventional animal agriculture could continue to expand.

Consumers accept cultivated meat

For this section, we assume that AGI solves taste parity by producing cultivated meat that matches or exceeds conventional meat on flavour and texture (see above). We discuss other factors relevant to food purchasing habits: price, familiarity, and trust.

The answer to the price question depends on what kind of world AGI produces. In a world where incomes remain broadly comparable to today, consumers are unlikely to switch to cultivated meat unless it is substantially cheaper than conventional meat. This is a high bar, and it becomes harder to clear if AGI also improves conventional animal agriculture, driving down the cost of factory-farmed meat. 

Some AGI proponents envision dramatic income growth. Even in a scenario where conventional meat is much more expensive than cultivated meat, sufficiently high incomes could render the difference irrelevant—in a post-AGI world with very high incomes, a steak costing even hundreds of dollars could be negligible. In either case (both very cheap, or conventional meat more expensive), the question shifts from price to preference, and we have no strong reason to think preference favours cultivated meat by default.

Only a handful of cultivated meat products are currently available on the market, so we do not yet have good evidence of how consumers feel about cultivated meat once it's available and they can actually try it. The limited evidence available suggests a low acceptance of cultivated meat, with food neophobia, perceived unnaturalness, and food safety concerns playing a role (see here for a review). A recent meta-analysis found that food technology neophobia—resistance stemming specifically from the technological nature of cultivated meat rather than simple unfamiliarity with novel foods—was a strong predictor of rejection.

There is also a growing headwind in the form of concern about ultra-processed foods. Cultivated meat is vulnerable to being categorized alongside heavily processed products that consumers increasingly report trying to avoid. The GMO analogy is frequently invoked in alternative protein circles. A technology that is scientifically defensible can still face lasting consumer rejection if a food safety scare, a wave of bad press, or a durable public misunderstanding takes hold. If AGI compresses the supply side and cultivated meat products appear widely in supermarkets before the relatively slow normalisation process has played out, sceptics could gain an easy argument along the lines of "this is being pushed on us before anyone knows the long-term effects." As the GMO precedent shows, that argument does not need to be scientifically valid to be politically effective.

A potential antidote to this is that AGI could meaningfully improve how cultivated meat companies communicate with consumers—identifying resonant framings across cultural contexts, running rapid message testing, and creating personalized advertising. But AGI would be equally available to the conventional meat industry, which could have far greater resources (as it does currently). Hyper-persuasive marketing is not a one-sided advantage for cultivated meat.

While AGI could remove the price and taste barriers, how it influences social normalisation is uncertain. AGI may even work against it if supply outpaces trust.

All of the above happen in coordination, on a timeline that matters for animals

Science, manufacturing, regulatory and political approval, and consumer acceptance are broadly sequential bottlenecks. AGI arriving in 2030 does not mean that alternative proteins displace animal agriculture in 2031.

AGI could solve the science quickly, but manufacturing takes time to build out, regulatory approval is jurisdiction-by-jurisdiction and slow, political opposition may be hard to counteract or intensify rather than fade, and consumer normalisation plausibly cannot be rushed without risking backlash.

Even if you grant generous odds at each stage, a chain of individually plausible steps can still produce an unlikely outcome. Say there is a 95% chance that AGI solves the science, an 85% chance manufacturing scales affordably, a 50% chance regulators speed up meaningfully, a 40% chance political opposition eases, and a 70% chance consumers accept the product—the probability that all of these resolve in coordination drops quickly. Under those assumptions (treating them as roughly independent), the combined probability is around 11%^[6].

There is also a lock-in risk. If AGI makes people richer before cultivated meat is widely available and familiar, higher incomes could increase meat consumption before alternatives are ready to capture that demand. We already expect the number of farmed animals to grow substantially over the next decade. A post-AGI income boom without available alternative proteins could accelerate that growth further, entrenching conventional agriculture and making displacement harder.

Even granting that AGI could eventually help overcome each bottleneck, the time it takes to work through them sequentially means years of continued and potentially increased animal suffering.^[7]

The bottlenecks that AGI seems least likely to help with—regulation, politics, and consumer trust—could be the ones that take the longest to resolve. If we want highly competitive alternative proteins to be ready when AGI-level capabilities arrive, those bottlenecks need attention now.

Caveats:

• Most of the conclusions drawn here about what is and isn’t AGI-tractable are based on current trends. However, you might think that current patterns of AI capabilities and deployment are unlikely to hold in a post-AGI world. We focused on the former because that allows actionable recommendations to be made.
• Similarly, we don’t address scenarios in which AGI drastically reshapes institutional and political dynamics. A sufficiently capable AI might find creative strategies for regulatory reform or public persuasion that we can't currently foresee. Governments and agencies could be restructured, approval frameworks could be overhauled, and entirely new institutional designs could emerge that bear little resemblance to current processes. As above, we focus on existing institutional structures because they allow actionable analysis, but we acknowledge this is a limitation.
• Some readers may think we're drawing the boundary of "AGI" too narrowly. We're happy to discuss what level of capability would change our conclusions.
• If you believe that the arrival of AGI means ASI arrives soon after, most of our takeaways probably don’t hold.

AI usage note:

• All research and reading were conducted by me.
• Claude helped draft various sections of text after I provided a detailed structure. All Claude-written text was thoroughly reviewed and edited by me.
• Claude made the map graphic.

1. ^{^}

   Assuming that each bottleneck is independent, which they may not be (see the final section for more detail)

2. ^{^}

   Scaffolding is a 3D structure that guides animal cells to attach and organize into tissues that conventional meat is found as, like steaks.

3. ^{^}

   By “science side” we’re referring specifically to the technical problems associated with going from lab-scale to commercial-scale production. Biology, fluid dynamics, and thermal physics do not scale linearly, which means cells and other inputs do not behave the same in small and larger bioreactors.

^{^}

Agencies could of course adopt more AIs to handle this increased volume, but they can only reduce backlog so much if humans are still required to make the ultimate decisions.

^{^}

Now eight.

^{^}

This calculation treats the steps as roughly independent, which is imperfect. Some bottlenecks are positively correlated. For example, if the science works well, manufacturing becomes easier. But others may be negatively correlated—if manufacturing scales fast and cultivated meat becomes a credible threat, political opposition could intensify. On balance, accounting for these dependencies does not obviously result in a high overall probability.

^{^}

This assumes that AGI does not substantially improve the welfare of animals on conventional farms through other means (e.g. vastly better monitoring, welfare enforcement, or shifts in moral attitudes). That is a separate and important question, but beyond the scope of this post.

Show all footnotes