A great article, thanks for sharing. I discovered it after listening to you on the Moonshots podcast. I appreciate the comprehensive analysis and the fact that someone with industry experience is willing to challenge the popular narrative of our day with thoughtful insights.
Hey Alvin. I absolutely love this article and thank you for planting the stake. I do have a few concerns. Apologies, response is a bit long.
This is a powerful and genuinely important framework. What I love about Abundanism is that it rejects the false premise that AI’s value should flow to a single winner. The real opportunity is not winner-take-all AGI, but competition, diffusion, falling costs, and the broad distribution of the resulting abundance.
I love the idea of AI as a partner in governance. Properly designed, AI could make government radically more efficient, reducing the administrative cost of governing while improving service delivery and accountability.
Purpose engineering is equally important. A society in which work changes profoundly must develop new systems for meaning, contribution, creativity, learning, and community.
My concerns however are not with the destination, but with the institutional machinery required to get there.
First, and as you suggest, diffusion must remain central. Governments may understandably impose limits around genuine national-security, biosecurity, or safety risks. But broad restrictions on open models, affordable inference, and cross-border access could amplify scarcity rather than overcome it. Recent actions by Anthropic and Open AI seem to be headed in this direction; proposing regulatory caption to create an anti-competitive moat. Competition matters because it prevents a handful of firms or countries from controlling the intelligence layer of the global economy.
Second, many of the most painful scarcities in Western economies are not simply natural; they are institutional. Housing, education, health care, food, transportation, and energy have become increasingly inaccessible not only because of limited resources, but because of supply constraints, regulatory capture, financialization, and systems designed to preserve rents. Housing is perhaps the clearest example: a basic human need has become a speculative asset class, often directly at odds with affordability for citizens. Marc Andreesen frequently rails on this idea. While Marc is arguably one of the most positive, optimistic and visionary leaders of our time, he remains skeptical on this issue of manufactured scarcity.
That raises a difficult question for any abundance model: how do we dismantle the incentive structures that monetize scarcity? AI can lower costs and expand capacity, but it cannot by itself overcome zoning barriers, monopolistic practices, licensing constraints, bureaucratic inertia, land speculation, or institutions that directly benefit from maintaining shortage.
Third, Abundanism depends on institutions being willing to reinvent themselves. If governments, financial systems, health-care organizations, and educational institutions resist structural change, AI-driven abundance may be captured by incumbents rather than propagated broadly. In that scenario, technology would not create a peaceful transition; rather it could intensify inequality, and social instability. The dystopian risk is not simply AI becoming too powerful but rather becoming an amplifier of the dysfunctional elements of our economy.
Finally, I agree that Pigouvian automation taxes deserve serious consideration as a way to price social externalities but they are unlikely, by themselves, to finance a large-scale UBI. For example, a payment of US$3,000 per month to 100 million people would cost US$3.6 trillion annually—roughly half of total federal outlays. In a political climate that cannot even pass a an affordable housing bill, not sure it could digest such a big idea that would require redesigned public services, complete rethinking of foreign policy (war regime) whose powerful advocates consume a disproportionate share of the budget. This could happen eventually but not without major crisis and/or revolution.
What makes this an even harder sell is the current data. We can’t actually assume that job displacement automatically means net job loss. The latest Ramp and Revelio Labs analysis found that high-intensity AI adopters increased white-collar headcount by 10.2% over two years, with entry-level hiring rising 12%. This compared to flat headcount growth for low-intensity adopters of AI. The study is not proof of causality but it is a useful warning against a simplistic “AI replaces everyone” narrative.
The challenge is about ensuring that productivity gains create new opportunity, lower the real cost of living, distribute the accrued value more equitably, and strengthen human agency rather than merely further concentrating wealth.
Abundanism offers a compelling north star. The next step is an equally rigorous transition architecture capable of breaking artificial scarcity, preserving competition, reforming institutions, and ensuring that abundance reaches people before social trust breaks down.
Dan, I couldn’t agree more. And yes much needs to change in the institutions and motivations of our leaders. I’m working on a new essay on how the transition will happen and what changes are needed. Look forward to your feedback when that comes out. Happy 4th! 🙏🎇
1. Explain the assertion that training costs must go up.
2. Nothing in here on the environment. Exponential increases in production and living standards simply must involve increased demands on finite resources, even air itself to say nothing of water and everything else. Can't wish this under the rug.
Training costs scale quadratically with size of model, context window and training data. So as frontier models grow all these parameters, the cost for training new models rise quickly. Particularly as data center costs are rising quickly per GW over time.
Yes near term environmental issues are real, but the over building will stop soon as the market realizes there’s no payoff. And smaller models will work find at the edge on your home devices or on your phone. Meaning there’s really not going to be major issues in this area after the bubble pops. ;)
Fantastic piece Alvin! So much information here to process. AI hype feels like an intellectual contagion that overrides measured, thoughtful planning by individuals, companies, nations and our entire species. Informed public discourse is necessary to debate the real potential benefits and risks from abundant intelligence and the intersection it has with the political, economic, behavioral, and other complexities that will drive our future. Keep driving attention to this critical architecture of our future.
Alvin, I can't agree enough with what you present here, as well as the evidence that you have to back up your assertions. You use Cisco as an example of how the infrastructure providers of disruptive technological innovation are NOT the ones that reap the benefits of change; I can second this with my experience in telco. For the last thirty years, the phone companies of the world have invested billions in building out the word's mobile networks, then upgrading them every decade or so, from 2G to 3G to 4G to 5G and now (in the works) 6G. Yet each upgrade has brought fewer monetary returns to the telco industry, as mobile connectivity is increasingly commoditized. The huge growth stories have come from the services that were enabled by improvements in global connectivity: Uber, Airbnb, Netflix, and many more. The phone companies can only watch enviously as their customers shell out far more money to these services than they will ever agree to pay for the connectivity that makes them possible. As you point out, this is what also happened to Cisco in the building of the internet, and it's very likely it will happen again with AI. It's not the basic tool itself that will generate the new revenues, it's what new services are built on top of them that will change people's lives and make all the money. So yes, expecting the "AI creation" companies to capture all new future value in this space flies in the face of what has happened before, and the oases of cash that these companies are expecting are indeed likely to turn out to be an extravagant - and costly - mirage.
More important, I applaud your call for a new set of AI policy and governing bodies to ensure that the changes that will come - and they will be big - don't create unrest and revolution in places where a well-fed elite has lost sight of how their casual decisions are causing traumatic mayhem for the people in the other leg of the K. Is it possible to actually build these bodies so that they have accountability and force in this modern world? Pope Leo issued a similar (if less specific) call in his May 15 encyclical about AI, so you're in good company! Now it's up to us, to humanity, to save ourselves in time. We're not too well known for taking our eyes off the shiny bauble flashing at us from just up ahead in favor of making proactive, pragmatic decisions, but with thought leadership like yours, maybe we've got a chance.
The narrated audio version of this essay filled a perfect gap in my afternoon today during a long drive to an appointment in Jupiter, FL This thoughtful call to action should be required for all policymakers, corporate leaders and stakeholders, as well as community leaders and organizers. Thank you Alvin for sharing your insights in such an approachable and accessible way.
Well done Alvin. Super clear narrative, super convincing conclusion. I agree with your take, this over-enthusiastic passion in a new technology is the hallmark of American capitalism, which provides the blood and energy of its society, as well as the economic rollercoaster excitement. Well, I hope this 3.5% isn't sharpening their pitchforks, and those AI titans can come to realization before the village gets burned down.
Thanks for taking the time to read it. I wrote this to highlight the need for change before it’s too late. Hope the lab heads take the time you have to read it.
A great article, thanks for sharing. I discovered it after listening to you on the Moonshots podcast. I appreciate the comprehensive analysis and the fact that someone with industry experience is willing to challenge the popular narrative of our day with thoughtful insights.
Thanks for taking time to read it. There’s other pieces on the substack channel you may also enjoy.
Hey Alvin. I absolutely love this article and thank you for planting the stake. I do have a few concerns. Apologies, response is a bit long.
This is a powerful and genuinely important framework. What I love about Abundanism is that it rejects the false premise that AI’s value should flow to a single winner. The real opportunity is not winner-take-all AGI, but competition, diffusion, falling costs, and the broad distribution of the resulting abundance.
I love the idea of AI as a partner in governance. Properly designed, AI could make government radically more efficient, reducing the administrative cost of governing while improving service delivery and accountability.
Purpose engineering is equally important. A society in which work changes profoundly must develop new systems for meaning, contribution, creativity, learning, and community.
My concerns however are not with the destination, but with the institutional machinery required to get there.
First, and as you suggest, diffusion must remain central. Governments may understandably impose limits around genuine national-security, biosecurity, or safety risks. But broad restrictions on open models, affordable inference, and cross-border access could amplify scarcity rather than overcome it. Recent actions by Anthropic and Open AI seem to be headed in this direction; proposing regulatory caption to create an anti-competitive moat. Competition matters because it prevents a handful of firms or countries from controlling the intelligence layer of the global economy.
Second, many of the most painful scarcities in Western economies are not simply natural; they are institutional. Housing, education, health care, food, transportation, and energy have become increasingly inaccessible not only because of limited resources, but because of supply constraints, regulatory capture, financialization, and systems designed to preserve rents. Housing is perhaps the clearest example: a basic human need has become a speculative asset class, often directly at odds with affordability for citizens. Marc Andreesen frequently rails on this idea. While Marc is arguably one of the most positive, optimistic and visionary leaders of our time, he remains skeptical on this issue of manufactured scarcity.
That raises a difficult question for any abundance model: how do we dismantle the incentive structures that monetize scarcity? AI can lower costs and expand capacity, but it cannot by itself overcome zoning barriers, monopolistic practices, licensing constraints, bureaucratic inertia, land speculation, or institutions that directly benefit from maintaining shortage.
Third, Abundanism depends on institutions being willing to reinvent themselves. If governments, financial systems, health-care organizations, and educational institutions resist structural change, AI-driven abundance may be captured by incumbents rather than propagated broadly. In that scenario, technology would not create a peaceful transition; rather it could intensify inequality, and social instability. The dystopian risk is not simply AI becoming too powerful but rather becoming an amplifier of the dysfunctional elements of our economy.
Finally, I agree that Pigouvian automation taxes deserve serious consideration as a way to price social externalities but they are unlikely, by themselves, to finance a large-scale UBI. For example, a payment of US$3,000 per month to 100 million people would cost US$3.6 trillion annually—roughly half of total federal outlays. In a political climate that cannot even pass a an affordable housing bill, not sure it could digest such a big idea that would require redesigned public services, complete rethinking of foreign policy (war regime) whose powerful advocates consume a disproportionate share of the budget. This could happen eventually but not without major crisis and/or revolution.
What makes this an even harder sell is the current data. We can’t actually assume that job displacement automatically means net job loss. The latest Ramp and Revelio Labs analysis found that high-intensity AI adopters increased white-collar headcount by 10.2% over two years, with entry-level hiring rising 12%. This compared to flat headcount growth for low-intensity adopters of AI. The study is not proof of causality but it is a useful warning against a simplistic “AI replaces everyone” narrative.
The challenge is about ensuring that productivity gains create new opportunity, lower the real cost of living, distribute the accrued value more equitably, and strengthen human agency rather than merely further concentrating wealth.
Abundanism offers a compelling north star. The next step is an equally rigorous transition architecture capable of breaking artificial scarcity, preserving competition, reforming institutions, and ensuring that abundance reaches people before social trust breaks down.
Dan, I couldn’t agree more. And yes much needs to change in the institutions and motivations of our leaders. I’m working on a new essay on how the transition will happen and what changes are needed. Look forward to your feedback when that comes out. Happy 4th! 🙏🎇
Alvin, only two things.
1. Explain the assertion that training costs must go up.
2. Nothing in here on the environment. Exponential increases in production and living standards simply must involve increased demands on finite resources, even air itself to say nothing of water and everything else. Can't wish this under the rug.
Looking forward to face time soon. B
Training costs scale quadratically with size of model, context window and training data. So as frontier models grow all these parameters, the cost for training new models rise quickly. Particularly as data center costs are rising quickly per GW over time.
Yes near term environmental issues are real, but the over building will stop soon as the market realizes there’s no payoff. And smaller models will work find at the edge on your home devices or on your phone. Meaning there’s really not going to be major issues in this area after the bubble pops. ;)
See you FTF soon.
Fantastic piece Alvin! So much information here to process. AI hype feels like an intellectual contagion that overrides measured, thoughtful planning by individuals, companies, nations and our entire species. Informed public discourse is necessary to debate the real potential benefits and risks from abundant intelligence and the intersection it has with the political, economic, behavioral, and other complexities that will drive our future. Keep driving attention to this critical architecture of our future.
Thanks for the feedback. Please share this with others that can benefit from reading it.
Alvin, I can't agree enough with what you present here, as well as the evidence that you have to back up your assertions. You use Cisco as an example of how the infrastructure providers of disruptive technological innovation are NOT the ones that reap the benefits of change; I can second this with my experience in telco. For the last thirty years, the phone companies of the world have invested billions in building out the word's mobile networks, then upgrading them every decade or so, from 2G to 3G to 4G to 5G and now (in the works) 6G. Yet each upgrade has brought fewer monetary returns to the telco industry, as mobile connectivity is increasingly commoditized. The huge growth stories have come from the services that were enabled by improvements in global connectivity: Uber, Airbnb, Netflix, and many more. The phone companies can only watch enviously as their customers shell out far more money to these services than they will ever agree to pay for the connectivity that makes them possible. As you point out, this is what also happened to Cisco in the building of the internet, and it's very likely it will happen again with AI. It's not the basic tool itself that will generate the new revenues, it's what new services are built on top of them that will change people's lives and make all the money. So yes, expecting the "AI creation" companies to capture all new future value in this space flies in the face of what has happened before, and the oases of cash that these companies are expecting are indeed likely to turn out to be an extravagant - and costly - mirage.
More important, I applaud your call for a new set of AI policy and governing bodies to ensure that the changes that will come - and they will be big - don't create unrest and revolution in places where a well-fed elite has lost sight of how their casual decisions are causing traumatic mayhem for the people in the other leg of the K. Is it possible to actually build these bodies so that they have accountability and force in this modern world? Pope Leo issued a similar (if less specific) call in his May 15 encyclical about AI, so you're in good company! Now it's up to us, to humanity, to save ourselves in time. We're not too well known for taking our eyes off the shiny bauble flashing at us from just up ahead in favor of making proactive, pragmatic decisions, but with thought leadership like yours, maybe we've got a chance.
Thanks Leslie. If we work together and apply the 3.5% rule, we still have a chance. We just need a more unified voice. Please share this with others.
The narrated audio version of this essay filled a perfect gap in my afternoon today during a long drive to an appointment in Jupiter, FL This thoughtful call to action should be required for all policymakers, corporate leaders and stakeholders, as well as community leaders and organizers. Thank you Alvin for sharing your insights in such an approachable and accessible way.
Thanks for taking the time to finish the longish piece. 😉 Please share with others that can benefit from its info.
Well done Alvin. Super clear narrative, super convincing conclusion. I agree with your take, this over-enthusiastic passion in a new technology is the hallmark of American capitalism, which provides the blood and energy of its society, as well as the economic rollercoaster excitement. Well, I hope this 3.5% isn't sharpening their pitchforks, and those AI titans can come to realization before the village gets burned down.
Thanks for taking the time to read it. I wrote this to highlight the need for change before it’s too late. Hope the lab heads take the time you have to read it.