Peter Thiel: The Stagnation of Science and the AI Revolution

Despite numerous challenges, the US remains relatively better positioned than other developed nations because it still generates innovative companies, with 11 of the 17 $100+ billion companies founded since 1990 being American versus 6 Chinese.

“companies uh started since 1990 that have Market capitalizations over 100 billion so brand new companies that have grown to being worth a hundred billion dollars or more there are 17 in the world um 11 are in the US six are in China zero everywhere else”

There are serious existential risks from nuclear weapons, AI, and environmental degradation that modern society has systematically downplayed or ignored due to safetyism and risk-aversion.

“I think there are sort of all these other dimensions of existential risk I think there's an AI dimension of existential risk that's uh you know where you know um you know I think La aers UD kowsky has gone kind of crazy but you know his arguments are not that bad and the people in Silicon Valley do not have great rebuttals to the uh existential risk of AI and there are you know and there are environmental issues not just climate change”

Inequality itself may be less important than stagnation: if GDP grew at 3% annually, people would be better off even if inequality remained large, because absolute living standards would improve.

“I always anchor a little bit more on the stagnation generally and so I think if you know I don't know if if the GDP grew at 3% a year um and even if the inequality was big and it was hard for people to move from Blue Collar to White Collar jobs um I think everybody would be better off”

American society is becoming politically polarized by ideology rather than by race or religion as it was in the past, though both political parties actually agree on most fundamental issues.

“we used to kind of sort Society Maybe by things like race or religion and today it seems much more like we're sorting by political ideology”

When Thiel was an undergraduate at Stanford in the late 1980s, majoring in any atoms-based field (mechanical engineering, chemical engineering, nuclear engineering, aerospace engineering) was widely considered a career mistake, while computer science was the only field offering real career prospects.

“when I was undergraduate stord the late 1980s uh it was a mistake to major in anything having to do with Adams mechanical engineering chemical engineering nuclear engineering AER asward terrible decisions and then the only thing that really worked were uh were were uh the world of bits computers internet mobile internet software”

Oil has highly inelastic demand: a 1% decrease in supply causes roughly 10% price increase, making it inherently prone to cartel behavior and collusion.

“oil again has this feature where if you increase um if you decrease supply of oil by 1% price goes up 10% so it has highly inelastic features 100 million barrels a day take a million barrels offline prices go up $10 a barrel”

PayPal struggled to expand beyond eBay into the broader offline payments market, which proved too difficult to cross, ultimately leading to the 2002 sale to eBay rather than continued independent growth.

“in the case of PayPal you know it it it felt at the end of the day like it was hard to radically expand Beyond eBay and so in 2002 we sold the company to to to eBay”

Yuri Gagarin, the first cosmonaut, died in a test flight decades after completing his historic orbit, exemplifying the risk-taking and casualties that were accepted in the space race.

“I think Yuri gagar and the first uh the first uh Cosmonaut who circled the Earth uh six seven years maybe a decade later died in some test fight test pilot”

The Marxist critique of cultural Marxism holds that when left-wing movements shifted focus from class and economics to race and gender, they abandoned the real economic problems, allowing elites to avoid addressing class inequality and housing costs.

“cultural Marxism um the Marxist critique of cultural Marxism is that it people to when they when they when they started focusing on race and gender they forgot about class they forgot about the real economics”

Simulation theory became psychologically attractive to AI researchers because it offered a solution to the unsolvable AI safety problem: if we're in a simulation designed by a superintelligence, then AI-human compatibility was already solved, so alignment need not be feared.

“the rough logic... was that uh if we're in a Multiverse um and you you have you build this Ai and um it's very and then there's a question as you build AI AGI super intelligence you know will it be safe... there were some pretty big theoretical problems... if it's smarter than you it might pretend to be friendly... so if your picture of the universe is a Multiverse then um it's you know the AI or the AGI is in the future... chances are we're all just going to die... if you're in a simulation theory of the universe where the simulation was designed created by some super AGI being then in some sense the AGI is in the past and the compatibility of AGI with humans seemingly was already solved”

The gap between the 50th percentile and 1st percentile in the US has not grown much, but the gap between the 1st and 0.01st percentile has grown substantially, indicating wealth concentration at the very top.

“the differences between let's say the 50th percentile and the you know the the the 1 percentile in the US is hasn't grown that much but the difference between the 1 percentile and the 0.1 or the 0.01 percentile has grown quite a bit”

Peer-reviewed science has become dogmatic and malthusian, functioning as a sclerotic institution that suppresses heterodox thinking and prevents scientists from being fired for orthodoxy.

“it's all sort of this uh corrupt peer- reviewed research it's incrementalism it's sort of a uh it's a stagnant malthusian sociopathic institution and uh and and to the extent The Sciences are the humanities are distracting us from The Sciences”

When oil companies adopt ESG policies and reduce production by 30%, the result can look like collusion with OPEC to raise prices and profits, though it's disguised as environmental virtue rather than antitrust violation.

“if all the CEOs got together in a room and said you know we are all going to cut our oil production by 30% and then the prices will go up by more than that and our profits are going to go way up um and we're in effect going to collude with OPEC and extend the OPEC cartel to all these companies um uh you know that would be a I don't know that would be a prima fascia violation of section two of the Sherman Act”

God and the question of God's existence is probably the biggest question there is, yet modern society downplays it to maintain peaceful coexistence, treating it as too dangerous to debate vigorously.

“the judeo-christian um part in particular um you know there's there's probably I I don't know sort of one of the maybe the biggest maybe the biggest thing you can talk about is God maybe God is the biggest thing there is and it's kind of a big difference whether or not God exists and um and uh and if we're in the society where where we want people to be uh um you know where we have we sort of um have peaceful coexistence by obscuring these big questions”

Wokeness and diversity debates function as a distraction from more important questions: economic stagnation, lack of scientific progress, religion and meaning, and housing affordability.

“I I I've come to worry that uh so many of these debates are distractions it's like a magic show where we're being hypnotized and we're paying attention to to it to a certain debate we're not seeing the the man in the orange monkey suit jumping on the back something like this and uh and that uh diversity is a diversion from more important things and the more important things can be questions of Economics questions of Science question of religion”

Nuclear weapons and the psychological realization around the early 1970s that humanity could destroy itself 20 times over shifted the culture away from funding bold science and toward regulation and constraint of scientists.

“there's something about the history of nuclear weapons where uh in my telling it it's a sort of delayed response it takes something like a quarter of a century for it to fully sink in but by by by the early 1970s it's like we can destroy the world 20 times over what are we doing does this make any sense and then you know maybe we shouldn't be funding the smartest physicists to build bigger bombs”

Science requires fighting a two-front war: against excessive dogmatism (as early modern science did against the Church) and against excessive skepticism. Current science fights only skepticism, making it as dogmatic as the medieval church.

“you're supposed to fight a two-front war against excessive dogmatism and against excessive skepticism... my my sort of rough qualitative sense if we fast forward to 2023 is that it is um 100% anti- skepticism”

New Atheism (Hitchens, Dawkins, Harris) in the mid-2000s was a politically correct way to attack Islam and Islamic terrorism, making it culturally acceptable to criticize religion indirectly.

“what they were doing in 2005 new atheism was a politically correct way to be anti-muslim and so you were uh you know God was this bad violent being um and uh and there sort of ways you made it all purpose you attacked all the different religions but the the real Target was you know Isis assama Laden”

New Atheism became irrelevant because by 2023 the geopolitical challenge shifted from Islamic terrorism to Xi Jinping and the CCP, but New Atheists have nothing to say about communist consensus, making their ideology obsolete.

“fast forward to 2023 you know the big geopolitical challenge for the West is not you know is not 7th Century fundamentalist Islam it's something like xingping thought and the CCP and then um and then this is sort of well it's sort of like a borg it's kind of this consensus thing where everybody believes what everybody else believes and it's disturbing the structure... the new atheists had some very powerful things to say against Bin Laden they don't have anything to say about uh why president G and the CCP are wrong and um and and so they are no longer relevant”

Decline in both religion and science in the US over the past 50 years is not coincidental, though the connection is not fully articulated—both reflect a shift toward avoiding dangerous big questions.

“you've also pointed out that we've seen both a decline in religion and science in the US over the last 50 years and on the face of it you know one would think maybe just a that that these things were opposed but you you have some sort of belief that they're linked”

Safetyism (excessive focus on safety) has spread from physical risk management to biomedical research, making it extremely difficult to conduct promising but risky medical research.

“on the you know biomedical side um uh which is you know an area that I've thought about a lot you know and always think and it always strikes me we could be doing so much more there's there's so many approaches that seem seem quite promising and uh and then um it is uh yeah the the the the barriers are just very very high and I think we are you we're scared of the things that can go wrong but we're not scared enough of the things that will go wrong if we do nothing”

Timeless eternal truths in advice (like 'don't kill people') are less interesting than historically specific guidance for what makes sense in 2023.

“I'm not interested in things that are Timeless Eternal truths I'm interested in figuring out things that are one time you know World historical what what what is it that makes sense for me or you to do in 2023 right here right now”

The American political situation is like Shakespeare's Romeo and Juliet: two families (Republican and Democratic) that are aristocratic and identical, fighting each other not because of meaningful differences but because of pure tribal rivalry.

“it's always the the Shakespeare versus Karl Marx Karl Marx people fight each other because they're truly different in Shakespeare they fight each other because there are no differences at all and so it's like the opening line of Romeo and Juliet um you know two houses Al like and dignity the monu and the Capulet and they hate each other and they're these two families that you know aristocratic families but they have no there's no difference at all”

Courage is in shorter supply than genius in modern society, meaning bold action and risk-taking are rarer and more valuable than raw intellectual capability.

“courage is in shorter supply than G genius is that new you think to our society or has that always been the case”

Venture capitalists should analyze monopoly defensibility in the same way companies do, seeking combinations of brand, network effects, economies of scale, or unique technology—but most VCs compete too much with each other to build genuine competitive advantages.

“as a VC you want to invest in companies that are unique that don't have uh you know too much uh competition nature red red and tooth and Claw but you also don't want to be competing too much with other VCS”

The Manhattan Project and Apollo program demonstrated that centralized government coordination and massive funding could accelerate science and technology, but this same centralization created large bureaucratic institutions that eventually slowed innovation.

“it was possible to accelerate Science and Tech through centralization and government funding... the Army was just telling people what to do it organized the scientists... they were able to bring this invention in three and a half short years... but then I think uh the longer term cost was that you created these these very large large bureaucratic institutions um you no longer had the innovations that were coming in that you could then scale”

Universities have become corrupted by wokeness in humanities curricula (English, history) while the real scientific problem—stagnant, incremental, peer-reviewed research and lack of innovation in STEM—remains ignored.

“the wokeness tends to be focused on the derangements of the humanities curricula and English or history or topics like that whereas uh whereas if the thing that's really wrong is that um the scientists aren't making any progress they're not inventing new things you know it's it's all sort of this uh corrupt peer- reviewed research it's incrementalism it's sort of a uh it's a stagnant malthusian sociopathic institution”

Almost all conventional wisdom and advice is bad because it is one-size-fits-all, and if advice works for everyone, it cannot be differentiating.

“I think almost all of it is because I don't know I don't know where to start I don't know where to start it's it's it's just uh it's just the real advice is it's I I just don't believe you're sort of in this one-size fits-all cookie cutter thing where everybody needs to do the same thing or you know if there's some conventional advice that works for everybody the fact that everyone's given it is not differentiating”

Pinker's argument that violence has declined is incorrect because he measures only kinetic energy (active violence) while ignoring potential energy (destructive capacity of nuclear weapons), which has increased dramatically.

“when you measure violence you're just measuring the kinetic energy you know how many bombs are being dropped yeah that's going down but the potential energy you know the number of nuclear bombs the ability you know the potential destructiveness is way higher than it was 50 years ago”

Safety regulations (seat belts, bicycle helmets, skiing helmets) provide real protection but represent a Laffer curve of safetyism where increasing stringency eventually becomes counterproductive; determining optimal level is difficult.

“there's some sort of like laugher curve of how far you go on the safetyism side right um yeah it's always it's always hard it's it's it's hard to articulate the you know the the kinds of places where it feels like we we've gone too far or where it's gotten hi you know it's sort of gotten hijacked by various rackets”

Modern heterodox thinkers in science are rare, found only in children's books (curious scientists exploring) or excluded from graduate labs; regimentation has replaced intellectual autonomy.

“maybe it's like a maybe it's in a children's science book where it's a little girl who's exploring the world and uh she is uh she's not dogmatic but that's you know it's it's in children's books we have the non-dogmatic scientists you know in grad school Labs or something like that where they're all regimented robots and we make sure that anyone who's uh even a little bit heterodox gets thrown off the overcrowded bus”

Nuclear deterrence theory relies on active fear and constant vigilance—it is not designed to work psychologically by forgetting about nuclear weapons; the fact that the US has not used nuclear weapons since 1945 may reflect exhaustion rather than the success of deterrence theory.

“maybe these nuclear weapons never get used um that's not the way nuclear deterrence was supposed to work was supposed to work you were supposed to think about them all the time and you're supposed to be scared and then you didn't use them and it sort of worked during Cold War from 1949 to '89 and and then the last you know 33 34 years it's sort of like we've just gotten psychologically exhausted from it we don't think about it any anymore but that's that's not really the theory of how this stuff is supposed to work”

We have been in a soft lockdown for approximately 50 years, not just during COVID, characterized by risk-aversion and bureaucratization that has prevented bold scientific and technological progress.

“we've been in a soft lockdown for for something like 50 years”

Innovation has been confined almost entirely to the world of bits (computers, internet, software) while progress in atoms (mechanical engineering, chemical engineering, nuclear engineering, aerospace) has been severely limited since the 1970s.

“it was a mistake to major in anything having to do with Adams mechanical engineering chemical engineering nuclear engineering AER asward terrible decisions and then the only thing that really worked were uh were were uh the world of bits computers internet mobile internet software”

For decades, total addressable market (TAM) has been overemphasized by venture investors, when in fact a very large TAM typically indicates heavy competition and a worse business opportunity than a smaller, defensible market that can be expanded later.

“but uh but there also are probably ways people can overstress on on something like uh like Tam and uh and I always think that if you have a tam that's too big um you know it obviously OBC hates other questions which may be more important if you have a very big total addressable Market you probably have a lot of competition and uh and then um and then um it that that that may create bigger challenges than you know um than having a a small Market or some something like that”

Europe represents an 'extraordinary failure' in innovation, unable to produce the scale of billion-dollar companies that the US and China generate.

“and then this is just the you know the extraordinary failure of of Europe all all these other places”

The US federal government's $32 trillion debt will eventually become a serious problem now that interest rates have risen above zero and are no longer below GDP growth rates.

“it seems it seems like a very big problem... at this point uh it it feels like finally some something broke we're you know no longer in a deflationary context uh um you know the interest rates have have spiked the inflation has spiked and uh I think we're heading for we're headed for a very very challenging decade”

Tobacco companies' profits increased dramatically after the 1990s settlement, as the agreement essentially cartelized the industry by preventing non-signatories from competing, turning them into a major tax revenue source.

“when the US government settled with the tobacco companies in the late 1990s um it effect calized the industry and then uh the government and the tobacco companies were on the side of massively raising prices and then it went from them being these terrible things to this major source of tax revenue and uh these uh these monopolies that just printed money like like crazy because no nobody was not part of the settlement could sell tobacco”

Thiel began articulating the technology stagnation thesis around 2008, though the core claim is that stagnation in physics, chemistry, and mechanical engineering has been occurring for roughly 50 years since the 1970s.

“I started uh talking about the tech stagnation probably around two 2008 uh and okay earlier and and the claim is that it's been uh it's been it's been slow for uh at this point running on five decades since since really since the 1970s”

The Stanford Review became significantly less heterodox and provocative starting in 2002, when the paper began posting articles online, because writers knew their words would be permanent and searchable.

“I started one of these uh conservative student newspapers at Stanford Stanford review back in 1987 and it was it was a wildly heterodox I maybe obnoxious... and you can date the exact point when it moderated and became much less that way it was 2002 it's when they started posting the articles on the internet”

Ten years ago most people in the AI community were militant atheists (Dawkins, Hitchens), but today most are creationists or simulation-theory believers, representing a significant cultural shift.

“10 years ago most of the people I knew in the AI Community were militant atheists but today it seems most of them are creationists they they think we're in a simulation”

Billionaires pay a lower effective tax rate (~15% via capital gains) than millionaires (50%+ ordinary income tax), creating a regressive tax structure where wealth concentration is taxed more lightly than income.

“I think the bli iones pay a lower tax rate than the millionaires the millionaires pay you know ordinary income tax you know 50% more of their income gets paid in tax... the billionaires it's let's say it's it's mostly capital gains taxes... maybe the effective tax rate for the billionaires is something like 15% five”

The LLMs and ChatGPT represent a breakthrough comparable in magnitude to the internet itself, marking genuine progress in artificial intelligence.

“I think the llms chat GPT you know it's it's probably a breakthrough that's you know I would I would rank as on par with the internet itself it's it is it is very big”

Interest payments as a share of GDP fell from 3.2% in 1991 to 1.6% in 2021, demonstrating how falling rates allowed debt to grow without immediate fiscal crisis.

“the annual cost of servicing the debt um in 2021 you was something like 1.6% of GDP whereas in 1991 it was 3.2% of GDP”

Social mobility decline is a real problem, but it becomes irrelevant in a zero or 1% growth economy where stagnation forces politics to be zero-sum and unresolvable.

“if we have zero or you know 1% GDP growth um you know um um it's it these things will be very very hard to solve and it'll be very Zero Sum very contentious and probably won't be solved”

Walmart in the 2000s rebranded itself as a 'green corporation' and focused on environmental initiatives (ESG) to split left-wing opposition, which was cheaper than paying workers higher wages, benefiting shareholders while avoiding underlying economic challenges.

“Walmart in the in the the 2000s was in always in the dogghouse because he wasn't paying its workers enough and they came up with the idea of rebranding themselves as a green Corporation and that sort of split the um uh the left-wing anti-walmart Alliance and then in effect um it was cheaper to do a little bit of green stuff then to pay the workers more”

When something is explicitly named as 'science' (political science, social science, climate science), it is likely not genuinely scientific—the naming convention itself signals insecurity and disciplinary weakness.

“when people name something science it's generally less scientific whether it's political science or social science Etc like how does that is that is that naming convention by itself also means science is help science progress less quickly... well it's it's sort of well it's always a qu it's it's sort of if you're insecure enough that you have to call something science then um yeah it's sort of like the gentleman do protest too much”

Computer science is a particularly ironic case: it has an inferiority complex relative to mathematics and physics, which is why it adopted the 'science' label despite being closer to engineering than natural science.

“computer science which was you know when I was an undergraduate at Stanford you know it was yeah it was it was the people who were not very good at e or math went into computer science and then um it was it was this it was much easier to get into than some of the other engineering you know and um and then um it had this inferiority complex and that's why the field uh um uh grabbed onto the science label”

The CCP under Xi Jinping represents a 'Borg'-like consensus system where everybody believes what everybody else believes, with a structure disturbing similar to East Bay rationalism and consensus scientific thinking, making it fundamentally different from Islamic fundamentalism and requiring different intellectual tools to critique.

“this is sort of well it's sort of like a borg it's kind of this consensus thing where everybody believes what everybody else believes and it's disturbing the structure of it not saying the content but the structure of it is disturbingly close to um the East Bay rationalists to uh to the sort of consensus scientific thinking”

Real estate/zoning in Los Angeles has gone too far on the safetyism side with extremely restrictive zoning laws; no new cranes on the skyline, indicating construction has stalled—an example of safetyism run amok.

“I think we've gone too far on the safety side with real estate where it's it's runaway nism y extremely strict zoning laws and I'm looking out of my window here in Los Angeles and it's all these Office Buildings that were built in the you know 60s 70s maybe 1980s are the most recent buildings I can't you know I I cannot see a single I can't see a single crane anywhere on the horizon it's just and that's that tells me we've gone way too far in something you know as relatively important as uh as real estate”

Founders Fund differentiated itself in 2005 by cultivating a brand as 'founder-friendly' and having founders as managing partners, giving it unique deal flow—but by 2023 this positioning has become more cluttered as others copy it.

“Founders fund had a brand where we are founder friendly we we started companies ourselves around the side of the founder that was a very differentiating brand and we started back in 2005 fast forward to 2023 lots of other people are saying it and so maybe maybe it's a little bit more cluttered”