Longest reported lifetimes, to scale
I've drawn six months as 183 days. The figures are from the papers and the company page you'll find in the sources below.
The short answer, by setting
| Setting | Longest reported | Where it comes from |
|---|---|---|
| Cortical organoids kept in a lab dish | Up to 694 days | Gordon et al., Nature Neuroscience, 2021 |
| Organoids wired to electrodes on the FinalSpark Neuroplatform | Up to about 100 days, more than 100 in some cases | Jordan et al., 2024 |
| Neurons on the chip inside a Cortical Labs CL1 | Up to six months | Cortical Labs (a company figure, and the cells grow flat on the chip, so it isn't strictly an organoid) |
I'd read the table with one thing in mind, which is that you aren't comparing like with like. The 694 days is a research group looking after organoids so they can study how they mature, and the other two are cultures doing work inside a machine, where the cells are also stimulated and recorded from around the clock. So if you're asking how long a living computer lasts, I think the honest answer today is a few months (maybe six at most).
The record in a dish: 694 days
I've found one figure that beats the rest, though I may have missed others. In a 2021 paper, groups at Stanford and UCLA grew cortical organoids from six stem cell lines (taken from five people) and kept them for up to 694 days, so you're looking at nearly two years. I think the question they cared about was whether organoids keep developing or get stuck at a fetal stage. Going by the epigenetic clock and gene expression, they'd reached stages that look postnatal at 250 to 300 days, which is roughly the timeline you'd see in a real brain.
If you're thinking about computing with organoids (the guide on what brain organoids are covers the basics), I'd say that matters. An organoid that's only a few months old is still very young tissue in developmental terms, and a 2019 study from the Muotri lab in San Diego found its electrical activity kept increasing over several months (with network events that resemble the EEG of preterm babies). So my guess is that an organoid you swap out every 100 days never gets to its most mature state, though I haven't seen anyone test that directly.
Why organoids die
You should know first that an organoid has no blood vessels. In a dish, oxygen and nutrients have to soak in from the outside, and a 2022 review by LaMontagne, Muotri and Engler puts the diffusion limit for oxygen at about 200 micrometres. Once an organoid gets thicker than that the middle starves (it's short of oxygen and glucose), and you end up with what's called a necrotic core, dead tissue at the centre that limits how far the organoid can grow and mature.
People have tried several fixes, and I'd expect more. You can grow blood vessel cells along with the neurons, or connect the organoid to a flow of fresh medium so the inside isn't left waiting for nutrients to diffuse in. The same review says vascularising them is still hard, because the vessels don't get far into the tissue and nothing flows through them yet. I wouldn't expect organoid lifetimes to jump until someone solves that, though I may be wrong.
Why a computer shortens the time
I couldn't find a paper that measures exactly why organoids on an electrode array die sooner than those in a dish, so I'll only say what's reported. FinalSpark's paper says that when the Neuroplatform started, its organoids lived for only a few hours, and that various improvements, mainly to the microfluidics feeding them, got that up to 100 days in the best cases. Over three years they replaced the organoids on their arrays more than 250 times. Cortical Labs builds life support into the CL1 and quotes six months for its neurons. You can see both are engineering numbers, and both still mean the culture dies and has to be replaced.
If you're running a computer on it, that's a real cost. Whatever a culture has learned goes with it, so the next one starts again from nothing, and the FinalSpark authors also note that an organoid's spontaneous activity changes over its lifetime. That's one reason I'd treat these systems as research tools for now. If you want the wider picture of what they can and can't do yet, the organoid intelligence guide covers it, and the biocomputing guide puts it next to DNA computing.
Questions
How long do brain organoids live?
I've seen up to 694 days for cortical organoids kept in a lab dish (Gordon et al., 2021). If they're wired to electrodes on FinalSpark's Neuroplatform, the best last about 100 days.
Why do brain organoids die?
They don't have blood vessels, so oxygen has to diffuse in from outside, and a 2022 review puts the limit at about 200 micrometres. If they grow past that you'll get a starved centre (it's called a necrotic core).
Do older organoids work better?
They're more mature, yes. In the 2021 study they reached postnatal-like stages at 250 to 300 days, and a 2019 study found their electrical activity kept rising over several months. I don't know of anyone who's tested whether that makes them better at computing though.
Sources
- Gordon et al., Long-term maturation of human cortical organoids matches key early postnatal transitions, Nature Neuroscience, 2021
- Trujillo et al., Complex oscillatory waves emerging from cortical organoids model early human brain network development, Cell Stem Cell, 2019
- LaMontagne, Muotri and Engler, Recent advancements and future requirements in vascularization of cortical organoids, Frontiers in Bioengineering and Biotechnology, 2022
- Jordan et al., Open and remotely accessible Neuroplatform for research in wetware computing, Frontiers in Artificial Intelligence, 2024
- Cortical Labs, CL1 product page