What began with an interesting, rather rosey discussion of inevitable changes in an industrial base, where tedious tasks become automated, as they invariably do and always have, then turned sharply into a narrow-minded America-rules-the-world bit of China-bashing. Very poor.
How does replacing outsourcing with automation address immiseration? In both cases the workers are made redundant. The way to change the trajectory is political, not technological.
I would be very suspicious of the figure (74-79%) for China’s share of the market. While it is certainly true that China has made great strides of late, both in terms of running early clinical trials and developing innovative drugs, this number is too high to be credible. That said, the US needs to up its game regarding certain parts of the drug development process-I am reasonably confident this will happen.
This is my area so I am more than familiar with the details. My criticism of the numbers does not invalidate the overarching thesis. Chinese companies such as Wuxi are embedded within the life sciences supply chain and many smaller therapeutics companies depend on them. However there are initiatives to mitigate this dependency (excluding them is one option although not optimal). Innovating your way through the problem is clearly superior.
I believe that such narrow specialization and niche tunnel experience will ultimately play a cruel joke with those who will implement it. As a result, we will get armies of "inept" theorists who will not have the slightest idea about the essence of the method, principles and nuances, will not be able to use basic tools and will be dependent on automated systems that will only give out data.
This is already a reality in many other fields. Anyone has to dip their hands in shit if they really want to become a researcher.
Another question, regarding the essence - why do we need all these systems? What benefit do they bring to the majority? What "magic" technologies have helped the world become safer, more peaceful, healthier or more prosperous over the past 5 decades? I think I can count the fingers on one hand. I remember the times when a doctor, after a detailed examination and simple tests, could make a fairly accurate diagnosis. Today, even with high technology and analysis, doctors make mistakes. The time when we hand over everything to machines except for the delivery of results will be the time of the final degradation of the human
Very interesting post. I would only be more cautious about one point: the characterization of Chinese biotech as having mainly scaled the “low-automation / high-flexibility” model, with “armies of PhDs performing manual benchwork.”
That may well be true in some outsourced biotech research and manufacturing segments — for instance in contract research organizations (CROs), which perform research, testing, preclinical or clinical work for other companies, and in contract development and manufacturing organizations (CDMOs), which help develop and manufacture drugs, biologics, or active pharmaceutical ingredients on behalf of biotech or pharmaceutical firms. But as a general description of Chinese biotech, the picture seems too one-sided and risks relying on an outdated view of China’s technological capabilities.
There are already several Chinese examples of highly automated or autonomous lab platforms. XtalPi describes “hundreds of automated lab stations and AGV vehicles”; ABB and XtalPi have partnered to build intelligent automated laboratory workstations in China; iBioFoundry in Hangzhou integrates robotic arms on rails for synthetic biology workflows; and Tsinghua / MegaRobo / Deepwave’s iAutoEvoLab has been presented as an industrial automated lab for programmable protein evolution.
So perhaps Ginkgo should not be framed as the automated future against a still-manual Chinese model, but rather as the U.S. expression of a broader global shift — one in which China is also already deeply engaged. This does not weaken the strategic importance of biotech, but it does qualify the U.S./China contrast.
The biggest problem in the world: the outlaw empire. Without the imbalance, for instance rare earth minerals, the empire will never be forced to stop subjugating and exploiting (colonization) of other countries. The USA financed 2 world wars (NYE committee 1934-1936, Standard Oil and other banks and corporations supporting the German Reich, the Berger committee in 1996, BIS: bank of international settlements), the militarization of the cold war (NSC-68) and many overt and covert operations around the globe, destroying the lives of hundreds of millions.
That has to end. So yes, it’s good that we have Iran, China and Russia being able to standup and be part of a multipolar world.
I think there are a lot of people not bothering whether the work is boring or repetitive, as long as it pays the bills. It’s a big over estimation that everyone can execute and want a challenging work environment and automation eventually kills the job market. Millions of people unemployed. And then what?
Peter, your models correctly identify Popular Immiseration as a primary driver of state collapse, yet your Substack analysis under-emphasizes how modern automation acts as an artificial accelerant to this exact variable. For centuries, immiseration occurred when human population growth outpaced agricultural land capacity, driving down the value of labor. Today, hyper-automation accomplishes the exact same result, but at warp speed. By eliminating predictable, bill-paying jobs, automation creates an artificial labor surplus. It strips the working class of their only leverage—their labor—driving their economic value to near-zero. Therefore, hyper-automation does not liberate humanity; it is the ultimate systemic engine of popular immiseration, mathematically ensuring the explosive growth of political instability.
I will add here my own text from 2024 on the same topic. Biotechnology is not only a matter of national security. It is also a potential instrument of global control — especially with the transition to personalized medicines tailored to each individual patient.
//
While some experts here are already burying the West and imagining a bright future for BRICS, the main technological breakthroughs are still happening in the West.
For example, EvolutionaryScale was a new biotechnology company whose main goal was to use artificial intelligence to predict protein structures, with broad potential applications in drug development and other biotechnological fields. (Note from today: in November 2025 the EvolutionaryScale researchers joined Chan Zuckerberg Biohub, where this line of work continues.)
At the time, they had released ESM3, the first generative model for biology capable of simultaneously analyzing the sequence, structure, and function of proteins. This model was trained on data covering an enormous diversity of natural proteins, which makes it unique and powerful. ESM3 can significantly accelerate processes such as the development of new drugs and biotechnological solutions for environmental restoration. This technology gives scientists the ability to create and study new protein structures more efficiently and more accurately, which may lead to major advances in medicine and other fields.
Imagine that you have a Lego set, but instead of blocks, you use molecules and proteins to build different things. This is what EvolutionaryScale helps scientists do: it helps them build these molecules and proteins much faster and more accurately, using AI that understands how they work. The ESM3 model learns from a huge amount of data about proteins and molecules found in nature. Thanks to this AI, trained on advanced chips, scientists can predict what new proteins will look like and how they will work. It is like a crystal ball that shows what shapes and functions new proteins may have even before they are created in a laboratory.
What will this give science? It will help scientists find cures for diseases faster, develop new treatment methods, improve biotechnologies for cleaning the environment, and create new materials. In the past, this could take years, but now it can be done in a matter of months or even weeks.
For example, if Alexander Fleming, who first isolated penicillin, had had access in his time to a program like ESM3, his research could have advanced much faster and more effectively. Penicillin itself is a small molecule rather than a protein, so a model like ESM3 would not design the drug directly — but it could work on everything around it: the proteins the antibiotic acts upon, the bacterial enzymes that break it down, and the targets that determine how resistance develops. Here is what it could have given him:
Fast identification of targets: The program could have helped Fleming quickly find and characterize the bacterial proteins and enzymes involved in how the antibiotic works — and in how bacteria defend against it. This would have reduced the time spent on experiments and testing.
Prediction of protein structures: ESM3 could have predicted the structures of these bacterial proteins, helping scientists understand at the molecular level why penicillin works, how resistance arises, and how the effect could be improved.
Design of new antibacterial strategies: The program could have suggested new antibacterial proteins, enzymes, or molecular targets, opening a path to therapies beyond what classical antibiotics alone could offer.
These new technologies will take medicine, and not only medicine, to a new level and will significantly accelerate everything. This will greatly accelerate the arrival of personalized medicines on the market — medicines made for a specific patient.
Thanks to ESM3, it will be possible to do the following:
Analysis of genetic information: The program can analyze a patient's DNA sequences in order to understand exactly which proteins or molecules may be most effective for treating that patient's specific disease.
Prediction of interactions: ESM3 can predict how different proteins and molecules will interact with the patient's body. This will help create medicines that are as effective as possible and have minimal side effects.
Fast development: The program can speed up the drug development process by suggesting new molecules and compounds that may be used for treatment. This is especially important for rare or complex diseases, where standard methods may not be effective enough.
Personalization of treatment: Using data about a specific patient, ESM3 can help create medicines that are ideally suited to that patient's unique biology. This may include both the dosage and the form of the medicine.
We are entering a world where the emphasis will be on the personalization of medicines according to the needs of each individual patient.
This is not only a major advance in medicine, but also a major opportunity for the intelligence services of a country that possesses these technologies to recruit the people they need around the world to their side. When medicines are standard and mass-produced, it is difficult to establish strict control. But medicines produced on the basis of a patient's DNA would make everything much simpler.
To this end, intelligence services would have every incentive to build DNA libraries — not so much on heads of state, who are specifically protected against exactly this, as on the far larger pool of people with access to state secrets: officials, officers, engineers, their family members. Obtaining someone's DNA is not particularly difficult when you have access to the person: a glass, a cup, a cigarette butt, cutlery in a restaurant, any object they have touched can serve as a source. There are simply too many such people, and they are far less protected than the top leadership.
Once the few companies that create personalized medicines were connected to such a database, intelligence services would gain effective control over this segment of the drug market. It would no longer be possible, for example, for some official from Russia to secretly order, through third parties, advanced medicines for serious diseases for himself or for his family.
Suppose some official they need, from another country, wants to order — for himself or for a member of his family — a medicine made on the basis of his DNA. In such a system, the moment his DNA is entered to create that personalized medicine, it could match an entry in the database, and the intelligence services would learn that a particular official from a particular country is ordering medicines for himself or his close relatives. And that is it: this official can now be turned into leverage. For the sake of his own health and the health of his loved ones, he would be willing to go very far.
But how would countries that do not have such breakthrough technology for personalizing medicines fight against this?
How could they protect themselves from the recruitment of their citizens who order personalized medicines abroad?
For this, they would create their own database of all officials and people who have access to state secrets, and connect it to all medical institutions in the country. As soon as an official or his relative went to a local hospital to take tests, and a serious disease was found, the database would flag this person and surveillance would begin. If the patient's condition improved more than it should have with the existing available medicines, this person would be placed under much closer control, because most likely he had gained access to medicines that are not available in the country. It would then become necessary to find out how he gained access and what he did to obtain it.
Dr Turchin, I’ve only recently become aware of your work, by way of the phrase “elite overproduction”, one of those clarifying descriptions that make an operating phenomenon suddenly visible to everyone. Even more interesting as a diagnosis of our own period is that it is a commonplace to all civilizations as they complexity themselves into cultural sclerosis.
What I’m wondering about with this little side excursion to the bio lab is what you bring up about labor replacement. It seems to me America is going to need a rejiggering of what we consider capitalist organization and maybe even theory. Ghettoizing every last bit of human work for the sake of fewer and larger bottom lines looks like high tech feudalism to me—and a massive step toward mass immiseration. Are you optimistic that this major phase transition can eventuate in a more widely prosperous society than the lame offer g held out for UBI for the plebs and Empire wealth for elites?
What began with an interesting, rather rosey discussion of inevitable changes in an industrial base, where tedious tasks become automated, as they invariably do and always have, then turned sharply into a narrow-minded America-rules-the-world bit of China-bashing. Very poor.
How does replacing outsourcing with automation address immiseration? In both cases the workers are made redundant. The way to change the trajectory is political, not technological.
I've seen highly automated systems for Elisa assay work in China.
I would be very suspicious of the figure (74-79%) for China’s share of the market. While it is certainly true that China has made great strides of late, both in terms of running early clinical trials and developing innovative drugs, this number is too high to be credible. That said, the US needs to up its game regarding certain parts of the drug development process-I am reasonably confident this will happen.
You should dig a little deeper into the matter before dismissing the statistic.
This is my area so I am more than familiar with the details. My criticism of the numbers does not invalidate the overarching thesis. Chinese companies such as Wuxi are embedded within the life sciences supply chain and many smaller therapeutics companies depend on them. However there are initiatives to mitigate this dependency (excluding them is one option although not optimal). Innovating your way through the problem is clearly superior.
OK.
I believe that such narrow specialization and niche tunnel experience will ultimately play a cruel joke with those who will implement it. As a result, we will get armies of "inept" theorists who will not have the slightest idea about the essence of the method, principles and nuances, will not be able to use basic tools and will be dependent on automated systems that will only give out data.
This is already a reality in many other fields. Anyone has to dip their hands in shit if they really want to become a researcher.
Another question, regarding the essence - why do we need all these systems? What benefit do they bring to the majority? What "magic" technologies have helped the world become safer, more peaceful, healthier or more prosperous over the past 5 decades? I think I can count the fingers on one hand. I remember the times when a doctor, after a detailed examination and simple tests, could make a fairly accurate diagnosis. Today, even with high technology and analysis, doctors make mistakes. The time when we hand over everything to machines except for the delivery of results will be the time of the final degradation of the human
Dear Peter,
Very interesting post. I would only be more cautious about one point: the characterization of Chinese biotech as having mainly scaled the “low-automation / high-flexibility” model, with “armies of PhDs performing manual benchwork.”
That may well be true in some outsourced biotech research and manufacturing segments — for instance in contract research organizations (CROs), which perform research, testing, preclinical or clinical work for other companies, and in contract development and manufacturing organizations (CDMOs), which help develop and manufacture drugs, biologics, or active pharmaceutical ingredients on behalf of biotech or pharmaceutical firms. But as a general description of Chinese biotech, the picture seems too one-sided and risks relying on an outdated view of China’s technological capabilities.
There are already several Chinese examples of highly automated or autonomous lab platforms. XtalPi describes “hundreds of automated lab stations and AGV vehicles”; ABB and XtalPi have partnered to build intelligent automated laboratory workstations in China; iBioFoundry in Hangzhou integrates robotic arms on rails for synthetic biology workflows; and Tsinghua / MegaRobo / Deepwave’s iAutoEvoLab has been presented as an industrial automated lab for programmable protein evolution.
So perhaps Ginkgo should not be framed as the automated future against a still-manual Chinese model, but rather as the U.S. expression of a broader global shift — one in which China is also already deeply engaged. This does not weaken the strategic importance of biotech, but it does qualify the U.S./China contrast.
Links:
https://en.xtalpi.com/lab_automation/
https://new.abb.com/news/detail/110608/prsrl-abb-robotics-partners-with-xtalpi-to-build-intelligent-automated-laboratories
https://synbioj.cip.com.cn/EN/10.12211/2096-8280.2023-027
https://www.nature.com/articles/s44286-025-00305-8
https://www.frcbs.tsinghua.edu.cn/scientific-research-results/1260
With respect to the 90% rare earth:
The biggest problem in the world: the outlaw empire. Without the imbalance, for instance rare earth minerals, the empire will never be forced to stop subjugating and exploiting (colonization) of other countries. The USA financed 2 world wars (NYE committee 1934-1936, Standard Oil and other banks and corporations supporting the German Reich, the Berger committee in 1996, BIS: bank of international settlements), the militarization of the cold war (NSC-68) and many overt and covert operations around the globe, destroying the lives of hundreds of millions.
That has to end. So yes, it’s good that we have Iran, China and Russia being able to standup and be part of a multipolar world.
I think there are a lot of people not bothering whether the work is boring or repetitive, as long as it pays the bills. It’s a big over estimation that everyone can execute and want a challenging work environment and automation eventually kills the job market. Millions of people unemployed. And then what?
Peter, your models correctly identify Popular Immiseration as a primary driver of state collapse, yet your Substack analysis under-emphasizes how modern automation acts as an artificial accelerant to this exact variable. For centuries, immiseration occurred when human population growth outpaced agricultural land capacity, driving down the value of labor. Today, hyper-automation accomplishes the exact same result, but at warp speed. By eliminating predictable, bill-paying jobs, automation creates an artificial labor surplus. It strips the working class of their only leverage—their labor—driving their economic value to near-zero. Therefore, hyper-automation does not liberate humanity; it is the ultimate systemic engine of popular immiseration, mathematically ensuring the explosive growth of political instability.
'Cultural evolution of humanity'.
Peter Thiel cs have a few ideas about that..🙂
Per the quasi-China bash....Biotech is another brick in the seeming actualization of Susan Strange's Theory of Structural Power over here.
Good post, Peter, and very informative.
I will add here my own text from 2024 on the same topic. Biotechnology is not only a matter of national security. It is also a potential instrument of global control — especially with the transition to personalized medicines tailored to each individual patient.
//
While some experts here are already burying the West and imagining a bright future for BRICS, the main technological breakthroughs are still happening in the West.
For example, EvolutionaryScale was a new biotechnology company whose main goal was to use artificial intelligence to predict protein structures, with broad potential applications in drug development and other biotechnological fields. (Note from today: in November 2025 the EvolutionaryScale researchers joined Chan Zuckerberg Biohub, where this line of work continues.)
At the time, they had released ESM3, the first generative model for biology capable of simultaneously analyzing the sequence, structure, and function of proteins. This model was trained on data covering an enormous diversity of natural proteins, which makes it unique and powerful. ESM3 can significantly accelerate processes such as the development of new drugs and biotechnological solutions for environmental restoration. This technology gives scientists the ability to create and study new protein structures more efficiently and more accurately, which may lead to major advances in medicine and other fields.
Imagine that you have a Lego set, but instead of blocks, you use molecules and proteins to build different things. This is what EvolutionaryScale helps scientists do: it helps them build these molecules and proteins much faster and more accurately, using AI that understands how they work. The ESM3 model learns from a huge amount of data about proteins and molecules found in nature. Thanks to this AI, trained on advanced chips, scientists can predict what new proteins will look like and how they will work. It is like a crystal ball that shows what shapes and functions new proteins may have even before they are created in a laboratory.
What will this give science? It will help scientists find cures for diseases faster, develop new treatment methods, improve biotechnologies for cleaning the environment, and create new materials. In the past, this could take years, but now it can be done in a matter of months or even weeks.
For example, if Alexander Fleming, who first isolated penicillin, had had access in his time to a program like ESM3, his research could have advanced much faster and more effectively. Penicillin itself is a small molecule rather than a protein, so a model like ESM3 would not design the drug directly — but it could work on everything around it: the proteins the antibiotic acts upon, the bacterial enzymes that break it down, and the targets that determine how resistance develops. Here is what it could have given him:
Fast identification of targets: The program could have helped Fleming quickly find and characterize the bacterial proteins and enzymes involved in how the antibiotic works — and in how bacteria defend against it. This would have reduced the time spent on experiments and testing.
Prediction of protein structures: ESM3 could have predicted the structures of these bacterial proteins, helping scientists understand at the molecular level why penicillin works, how resistance arises, and how the effect could be improved.
Design of new antibacterial strategies: The program could have suggested new antibacterial proteins, enzymes, or molecular targets, opening a path to therapies beyond what classical antibiotics alone could offer.
These new technologies will take medicine, and not only medicine, to a new level and will significantly accelerate everything. This will greatly accelerate the arrival of personalized medicines on the market — medicines made for a specific patient.
Thanks to ESM3, it will be possible to do the following:
Analysis of genetic information: The program can analyze a patient's DNA sequences in order to understand exactly which proteins or molecules may be most effective for treating that patient's specific disease.
Prediction of interactions: ESM3 can predict how different proteins and molecules will interact with the patient's body. This will help create medicines that are as effective as possible and have minimal side effects.
Fast development: The program can speed up the drug development process by suggesting new molecules and compounds that may be used for treatment. This is especially important for rare or complex diseases, where standard methods may not be effective enough.
Personalization of treatment: Using data about a specific patient, ESM3 can help create medicines that are ideally suited to that patient's unique biology. This may include both the dosage and the form of the medicine.
We are entering a world where the emphasis will be on the personalization of medicines according to the needs of each individual patient.
This is not only a major advance in medicine, but also a major opportunity for the intelligence services of a country that possesses these technologies to recruit the people they need around the world to their side. When medicines are standard and mass-produced, it is difficult to establish strict control. But medicines produced on the basis of a patient's DNA would make everything much simpler.
To this end, intelligence services would have every incentive to build DNA libraries — not so much on heads of state, who are specifically protected against exactly this, as on the far larger pool of people with access to state secrets: officials, officers, engineers, their family members. Obtaining someone's DNA is not particularly difficult when you have access to the person: a glass, a cup, a cigarette butt, cutlery in a restaurant, any object they have touched can serve as a source. There are simply too many such people, and they are far less protected than the top leadership.
Once the few companies that create personalized medicines were connected to such a database, intelligence services would gain effective control over this segment of the drug market. It would no longer be possible, for example, for some official from Russia to secretly order, through third parties, advanced medicines for serious diseases for himself or for his family.
Suppose some official they need, from another country, wants to order — for himself or for a member of his family — a medicine made on the basis of his DNA. In such a system, the moment his DNA is entered to create that personalized medicine, it could match an entry in the database, and the intelligence services would learn that a particular official from a particular country is ordering medicines for himself or his close relatives. And that is it: this official can now be turned into leverage. For the sake of his own health and the health of his loved ones, he would be willing to go very far.
But how would countries that do not have such breakthrough technology for personalizing medicines fight against this?
How could they protect themselves from the recruitment of their citizens who order personalized medicines abroad?
For this, they would create their own database of all officials and people who have access to state secrets, and connect it to all medical institutions in the country. As soon as an official or his relative went to a local hospital to take tests, and a serious disease was found, the database would flag this person and surveillance would begin. If the patient's condition improved more than it should have with the existing available medicines, this person would be placed under much closer control, because most likely he had gained access to medicines that are not available in the country. It would then become necessary to find out how he gained access and what he did to obtain it.
//
Is this related to the work that Lee Cronin is doing in England?
Dr Turchin, I’ve only recently become aware of your work, by way of the phrase “elite overproduction”, one of those clarifying descriptions that make an operating phenomenon suddenly visible to everyone. Even more interesting as a diagnosis of our own period is that it is a commonplace to all civilizations as they complexity themselves into cultural sclerosis.
What I’m wondering about with this little side excursion to the bio lab is what you bring up about labor replacement. It seems to me America is going to need a rejiggering of what we consider capitalist organization and maybe even theory. Ghettoizing every last bit of human work for the sake of fewer and larger bottom lines looks like high tech feudalism to me—and a massive step toward mass immiseration. Are you optimistic that this major phase transition can eventuate in a more widely prosperous society than the lame offer g held out for UBI for the plebs and Empire wealth for elites?