Bernie's plan sucks, actually
In a Nutshell
Bernie Sanders' proposal to seize 50% equity in AI companies for a sovereign wealth fund fails because it targets the wrong firms, creates constitutional and capital flight risks, and gives government active control that invites political abuse. A workable alternative would use broad ecosystem taxes like compute levies and windfall surtaxes, maintain passive ownership capped at 10% with no board seats, and distribute proceeds through automatic formulas including citizen dividends. This approach follows proven models like Norway's fund and Alaska's Permanent Fund while avoiding the incentive destruction and market distortions of forced confiscation.
These notes were generated by AI and may contain inaccuracies.
Bernie Sanders proposed capturing 50% of AI companies including OpenAI, Anthropic, and xAI through a one-time expropriation of wealth. The proposal was introduced via a New York Times op-ed on June 1st rather than through actual legislation. The mechanism involves a one-time 50% tax paid in company stock, giving the government equity and a controlling stake in these AI labs. The vehicle would be a newly created federal sovereign wealth fund.
Sovereign wealth funds are supported at multiple levels of government including federal, state, and municipal. However, the specific approach proposed differs significantly from established models. The moral foundation rests on the argument that generative AI models are trained on humanity's collective knowledge, creative work, and data without compensation, and that significant research was publicly funded.
A Senate office report estimates up to 97 million potential jobs displaced over a decade due to AI disruption. This figure comes from official government analysis rather than private sector estimates.
The first mandate calls for government to receive voting shares and equal board representation in the named labs, with power to block corporate decisions deemed harmful to citizens. The intent is to prevent a handful of executives from dictating societal transformation behind closed doors. This approach mirrors the Chinese Communist Party's control over major companies and creates significant risks of political capture. The government changes its positions frequently, and elected politicians should not have unilateral control over private enterprise.
The second mandate involves distributing dividends as direct cash payments to the American people, with long-term funding earmarked for healthcare, education, and housing. The Norwegian sovereign wealth fund provides up to 20-30% of total household compensation through in-kind services rather than direct cash dividends, effectively increasing household resources without direct cash transfers.
Free healthcare and free education are not fungible and therefore not necessarily the optimal target for wealth distribution. The economic concept of double coincidence of wants explains why direct cash payments are necessary alongside in-kind services. Housing and transportation require ongoing funding, while needs like healthcare and education vary over time. Universal basic computer or universal basic tokens would only benefit approximately 1-5% of people who can effectively use those resources, as AI will eventually become the dominant intelligence in economic competition.
Forcing firms to surrender half their equity destroys valuations and drives founders, engineering talent, and capital to foreign jurisdictions. Even mild wealth taxes have this effect, with companies relocating from California to Texas or re-housing data centers in Ireland. Highly mobile firms will move assets, resources, and talent to the most tax-friendly locations.
The takings clause of the United States Constitution would classify a one-time stock transfer with inadequate compensation as unconstitutional, inviting immediate paralyzing constitutional challenges.
Treating frontier innovation purely as oligarchic extraction ignores how risk, capital markets, and execution build value. Venture capitalists and investment bankers would rationally allocate capital away from investments carrying a 50% expropriation risk, even if alternative options are less exciting.
Norway's sovereign wealth fund limits holdings to no more than 10% of any particular security to avoid market distortion. Bernie Sanders' proposal explicitly seeks market-distorting power. The federal government would become the regulator, customer, owner, and beneficiary simultaneously. The state cannot neutrally regulate or buy from a company when it relies on that company's dominance to fund the national budget.
Even when wealth funds are intended to pass proceeds to citizens, administrations may redirect funds toward military-industrial complex priorities. National wealth funds require constitutional protection, as demonstrated by the Alaska Permanent Fund and Norway's sovereign wealth fund.
The proposal focuses exclusively on visible companies with high R&D cash burn such as OpenAI, Anthropic, and xAI, which are not cash flow positive. It ignores hyperscalers and enterprise companies including Microsoft, Google, and Amazon that are actually generating profits. It also overlooks compute and infrastructure winners including Nvidia, CoreWeave, TSMC, and ASML.
The proposal functions as a messaging bill or Overton window pivot. Starting with a hyperbolic, extreme idea of 50% forced equity seizure and board vetoes shifts the policy discussion from the status quo of zero public capture toward more moderate positions such as wealth funds with broader capture mechanisms.
Revenue should target ecosystem-wide AI rents rather than specific company names, capturing compute and extraordinary windfalls across the entire sector. Governance should remain passive, with the government stripped of board seats, holding wealth neutrally with a 10% cap as Norway does. Shares should be purchased on the open market with actual money rather than through forced transfers. Distributions should be automatic and formulaic, removing day-to-day political capture.
A confiscation model is fundamentally flawed. A gradual share levy with self-valuing through open market purchases avoids cash flow shocks. The Alaska Permanent Fund captures 25% of oil revenues as an ongoing rent from the commons. Multiple sovereign wealth funds exist beyond resource extraction models.
A frontier compute levy would impose a small tax on GPU hours at the data center level. Data center taxes already generate significant county revenue in places like Virginia. A windfall surtax of 15-25% could apply strictly to monopoly-style AI rents above normal historical benchmarks. Public warrants following a Chips Act model would share upside for firms using federal subsidies, power grid priority, or public land.
The Santiago Principles establish best practices for public wealth funds. Active control by government has failed in every historical example. Professional management through independent fiduciaries is the globally established model.
No board seats or voting shares should be granted to the federal government. Strict exposure limits should cap holdings at 10% of any single public company. Management should be delegated to independent third-party fiduciaries insulated from White House or congressional stock picking. The fund must be constitutionally protected to prevent future raids.
A recommended allocation includes 40% to an AI dividend with quarterly or annual payouts to citizens, 25% to worker transition programs, 15% to public infrastructure including schools, research compute, hospitals, roads, power grid, and natural restoration, 10% to local grid compensation rebating communities bearing power and water costs of AI data centers, and 10% for reinvestment.
Sanders' proposal targets narrow named labs through one-time 50% forced equity transfer with active board seats and veto power for government, creating high risk of capital flight. A realistic blueprint would target ecosystem-wide companies through compute levies, dilution, and windfall taxes with strictly passive index ownership. The realistic approach follows subsidiarity principles, allowing cities, counties, and states to participate in capturing rents from data centers within their jurisdictions to fund local services.
The speaker notes that their room is very small and echoey. They mention testing the audio setup and invite viewers to provide feedback in the comments if the audio quality is acceptable, in which case they will continue using this setup. The speaker describes themselves as a perfectionist who wants the audio to sound good, though they clarify they are not an audiophile.
The speaker thanks viewers for watching and expresses hope that the content was valuable. They mention posting a deeper dive in both their school community and on Patreon for those wanting more detailed analysis. Additionally, they will publish a blog post on Substack that provides further depth and will not be paywalled. The speaker closes by thanking the audience again and wishing them well.
Keep David Shapiro in your library
Save the videos and channels worth coming back to, and find them again in one place.
