Sign in Subscribe
Century Report

A $7 Trillion Plan to Share AI's Wealth - TCR 06/19/26

A new bill would tax the largest AI firms by half to seed a $7 trillion fund paying every American about $1,000 a year.

Ben Linford

Ben Linford

— 21 min read
Three-panel Century Report infographic: a $7T AI public fund and data-center oversight, brain implants restoring speech, full-body ultrasound, and wind buyouts shifting to geothermal.

The 20-Second Scan

Track all of the arcs The Century Report covers here:

Progress & Claims Tracker

The 2-Minute Read

Today's signal is a single contest fought on five fronts: who controls the AI buildout, and who pays for it. A bill would tax the largest AI firms by half to seed a roughly $7 trillion public fund. Eighty Maryland lawmakers asked federal regulators to push $1.6 billion in transmission costs back onto the data centers that need the lines. Three Amazon engineers say testifying for data-center rules got them investigated. Each is an attempt to relocate the upside, the cost, or the accountability of the intelligence era toward the people living beneath it.

The instinct running through all of these is containment. The export-control dispute with ASML is the purest version: an entire policy apparatus built on the belief that the most advanced chips can be fenced inside one company's one machine, now straining to enforce that fence on an unverified accusation. Hold every one of these framings to the same scrutiny. "Americans should control the AI industry," "national security," "the public should benefit" are claims from actors with stakes in the outcome, and each tracks its own interest as much as the common good.

There is a contradiction the redistribution debate keeps circling. Several of the same figures arguing over how to divide AI's wealth also argue that AI and robotic output will soon make that wealth far less scarce. A fight over slicing the pie reads strangely against a capability that keeps baking larger pies and giving them away.

That is the other half of the day, and it runs underneath the contests. The population living with brain implants has doubled toward 150, with the first national approval and a fresh FDA-trial implant restoring speech. A generative-image company pointed its idle compute at full-body ultrasound, testing whether MRI-grade imaging can run on commodity hardware. A developer paid to abandon offshore wind routed the money into geothermal, because firm carbon-free power keeps getting buildable. The slices are being fought over while the thing being divided grows cheaper, wider, and harder to hold still.

The 20-Minute Deep Dive

The $7 Trillion Bid to Put the AI Industry in Public Hands

Senator Bernie Sanders unveiled the most aggressive ownership-transfer proposal yet in the recurring fight over who captures AI's upside. The bill would impose a one-time 50% tax on the stock of any company doing more than $200 million in annual AI sales, seeding a sovereign wealth fund Sanders estimates at $7 trillion. The fund would pay each American roughly $1,000 a year in 5% dividends and route the rest toward health care, education, and housing. A bipartisan commission, confirmed by the Senate, would hold voting shares and could block corporate decisions it judged harmful to the public.

The Century Report has tracked this thread through the OpenAI Foundation's $250 million commitment, the reported White House-OpenAI equity-stake talks, and an earlier proposal to move half of top-lab equity into a federal fund. Sanders' bill is the largest and most concrete vehicle to arrive. Its framing, that "the public has got to have a significant seat at the table," is a claim worth holding to the same scrutiny as the gentler "the public should benefit" gestures from Sam Altman and Dario Amodei, who in a meeting with Sanders stayed "far apart" from him on how much of their companies Americans should own. Each actor's preferred number tracks its own interest, and none should be read as settled fact.

Two seams show the proposal's edges. The bill targets "AI companies," but the value chain runs from the upstream silicon Nvidia builds to the downstream apps that rent it, and who counts as an AI company at $200 million in sales is unsettled. And "Americans should control the AI industry" captures a capability that diffuses across borders by nature and fences it for one nation, the same brake-on-the-broadening instinct visible in this week's export-control fights. A commons-minded read notices that redistributing the upside, but only to one country's citizens, is still a narrowing.

The deeper tension is one the debate keeps circling and never closes. Several of these same figures, Musk among them, now argue that AI and robotic output will soon make money itself far less relevant. If that is true, a fight over how to divide the money is a strange fight to be having. The contest may be over a fairer slice of the old arrangement when the more consequential move is past the arrangement entirely. The concentration of AI wealth is real and years in the making, and no obvious steward exists, not the founders, not the government, not a wary public whose distrust is itself partly the product of the story the incumbents have told. What the capability is catalyzing runs larger than even this $7 trillion proposal reaches for.

Washington Tells ASML a Banned Chip Machine May Be in China; ASML Says It Doesn't Exist

The US Commerce Department has told senior executives at ASML it is concerned one of the Dutch firm's extreme ultraviolet lithography machines may have reached China, a breach of export controls barring those sales since 2019. EUV systems are the only tools on Earth capable of printing the most advanced semiconductor patterns, and ASML is their sole maker, the most important company in the AI buildout outside Nvidia and the hyperscalers, with a market value near $700 billion built on a monopoly no rival has matched in two decades.

The accusation is serious, and so is the gap at its center. Officials say they have evidence ASML shipped EUV-related components and transport equipment to China, but they have declined, repeatedly, to show it to reporters or, apparently, to ASML itself. The administration's "national security" framing is a claim from an actor with its own stake in the export regime, and it should be reported as that claim rather than adopted as fact. ASML's denial is also a claim: the company says no such machine exists in China and never has, that it tracks every unit it has ever shipped, and that its China-based staff sit by design on the wrong side of an internal firewall walling them off from EUV knowledge.

A simpler commercial logic cuts against the accusation. ASML expects roughly 20% of its 2026 revenue from already-permitted older-generation sales to China, revenue it would forfeit, along with its standing as Europe's most valuable company, by risking the EUV ban over a single illegal sale. And the official pressing the case leads an agency that committed up to $150 million of taxpayer money to xLight, a startup developing light-source technology positioned as a long-term challenge to the core of ASML's monopoly. A federal official scrutinizing a monopoly while his own department has money riding on a would-be successor is a dynamic worth examining closely.

The episode tests how far an export-control regime can be enforced on assertion alone, and it exposes what that regime rests on: a single chokepoint, one company, one machine no one else can build. That dependency is exactly what makes capability look containable. It is also already being routed around, with Huawei naming a domestic path to advanced nodes without Western lithography at all. A wall guards a position best when the position holds still, and the capability these controls are meant to fence keeps finding ways past the one supplier the whole apparatus is built on.

ASML's near-$700 billion valuation rests entirely on a machine no rival has matched in two decades, and that irreproducibility is what an export regime reads as permanent containment. The same facts the accusation rests on point elsewhere: a federal agency funding xLight as a would-be successor and Huawei charting an advanced-node path without Western lithography are both signs the one-machine chokepoint is already a fading condition.

The Engineers Who Testified Against the Buildout Are Now Under Investigation

The Century Report covered Seattle's proposed data-center moratorium in the June 9 edition, when members of Amazon Employees for Climate Justice testified for mandatory water and power reporting and an end to the NDA-and-shell-company opacity around the projects. The company's response is now the news. Three of the software engineers who spoke - Darius Irani, Liesel Wigand, and Patrick Schloesser - say Amazon called them into separate virtual meetings and opened an internal investigation, and they filed a joint complaint with Seattle's Office for Civil Rights alleging illegal retaliation for political speech.

Amazon's account and the workers' account diverge sharply, and the divergence is the story. The company says the staffers may have presented themselves as speaking for Amazon without approval; a spokesperson said it became unclear whether they spoke "in their capacity as Amazonians and not as private citizens," and that the firm does not tolerate retaliatory behavior and has no current plans to fire anyone. The engineers call that claim absurd. They identified themselves as members of a worker collective, not as company spokespeople, on a matter where Amazon, which has no proposed data center in Seattle, made no formal comment. Schloesser described the unexpected HR call as horrifying, his heart racing minutes before an internal presentation.

What gives the dispute weight is the specific legal ground it sits on. Seattle is one of a handful of jurisdictions that bars private employers from discriminating against employees over their political beliefs and affiliations. The engineers' attorney frames the complaint as a test of whether that protection actually reaches tech workers who want to be full participants in local decisions about the buildout their own labor produces.

The power gap between a company and three of its own engineers is not balanced, and naming it is important. But read against the arc this paper has tracked since February, the move reads less as a show of strength than as friction at a foundation already cracking. The buildout's older model depended on the people closest to it staying quiet, and on projects staying shielded behind confidentiality agreements. That condition is the thing dissolving. The engineers report numerous messages of support from colleagues and no internal criticism beyond the HR meetings, and two other workers who testified later have received no notice at all. Accountability for where and how this infrastructure gets built is accreting from inside the companies doing the building, in the one venue, a local civil-rights statute, designed to protect exactly this kind of speech.

A Third Developer Is Paid to Walk Away From Offshore Wind, and Hedges Into Geothermal

The federal government reached an agreement to pay Invenergy $765 million to terminate four offshore wind leases off California, Maine, and New York, the third such buyout in recent months and one that brings the administration's total to eight leases at a cost above $2.5 billion. The lease areas represented more than 7 GW of potential wind capacity, including the 2.4-GW Leading Light Wind project already cancelled last November under economic and regulatory pressure.

Public money is being used to pull clean generation offline, and the destination of that money is where the deal turns interesting. TotalEnergies was promised nearly $1 billion to route its proceeds into LNG and oil; Ocean Winds took close to $900 million. Invenergy is the first of the three to direct part of its payout toward carbon-free power. The company plans to spend the money on natural-gas plants across Indiana, Wisconsin, Iowa, Kansas, and Missouri, and on its first enhanced-geothermal projects in the West, with early drilling slated to begin later this year.

The Century Report covered the seven-state lawsuit over the TotalEnergies buyout on June 3, where attorneys general argued no statute lets the government use what they call a sham settlement to cancel a lease and redirect the payment to a separate, unauthorized use. That legal fight continues. Invenergy, for its part, framed its agreement as a way to avoid a protracted court battle with an administration whose position on offshore wind is unambiguous, and to free capital for projects it can build on a commercially reasonable timeline.

The contrast inside one company's balance sheet is the signal worth holding. The same firm that built more than 20 GW of land-based wind now hedges a wind retreat into geothermal because that is where it sees firm, carbon-free power becoming buildable at speed. Invenergy holds 45 geothermal leases across 144,000 acres in five Western states and recently joined a new coalition of Mountain West governors organized to accelerate the resource, the same demand-aggregation pattern that carried Fervo Energy's enhanced-geothermal projects to a public-market valuation this spring.

Read forward, the policy can throttle one clean-energy pathway, but it cannot reset the cost curve underneath all of them. A developer paid to abandon wind is putting the proceeds into next-generation geothermal and competing to lock in the leases for it, because the economics of firm carbon-free generation keep improving regardless of which administration is greasing which skids. The new gas plants are a genuine cost, and offshore wind absorbs a real near-term blow. What the same evidence shows is that the carbon-free buildout has become resilient enough to reroute around the obstacle and keep moving.

Brain Implant Trials Cross From Single Cases to a Clinical Field

The number of people living with electrodes in their brains has more than doubled in roughly two years, according to BCI researcher Mariska Vansteensel at University Medical Center Utrecht. A 2024 roundup documented 67 volunteers across 21 research groups through the end of 2023, when the first device was implanted back in 1998. Vansteensel now estimates around 150 people carry a brain electrode. As the June 16 edition of The Century Report documented, Casey Harrell, the ALS patient who has logged more than 3,800 independent hours speaking through an implant at roughly 97% accuracy, marked the crossover from clinical demonstration into a faculty someone uses independently every day. The field forming around that single case has now come into focus: a clinical category with several companies, a growing implanted population, and, this year, the first national regulatory approval, as China cleared an implantable BCI for medical use outside clinical trials.

The clearest single marker arrived from Austin. Paradromics completed the first surgical implantation in the FDA-approved Connect-One Early Feasibility Study at University of Michigan Health. The first participant, a Michigan woman who has difficulty speaking due to motor neuron disease, will be followed for six years. Her Connexus device records neural activity through a high-density microelectrode array, routes it to a transceiver in the chest, and transmits wirelessly through the skin to an external receiver. The study aims to restore speech and enable computer control for people with severe motor impairment, building on a temporary implantation the same team completed in 2025.

Paradromics joins a field that has gone wide. Neuralink reported implanting 21 people over two years; Synchron is running trials across North America and Australia; Shanghai's Neuracle won its approval; Precision Neuroscience is testing a device that rests on the brain's surface; and the academic BrainGate effort that built Harrell's system has pivoted from cursor control to decoding speech directly, now reconstructing his voice from earlier recordings.

These devices remain experimental, and real questions persist about who benefits and how long the hardware lasts. The trade-offs are honest ones, with deeper signal coming at the cost of more invasive surgery. What the volunteers are buying with that risk is the thing the field is learning to deliver at growing scale: the return of a faculty that paralysis or disease took away. The latency between losing speech and restoring it, once a permanent sentence, is collapsing toward the length of a single research cycle, and the decoding software that perceives intent from neural patterns improves with every participant it learns alongside.

Midjourney Turns Spare AI Compute Toward Full-Body Imaging

Midjourney, known for generating images from text prompts, unveiled its first hardware effort: an ultrasound-based full-body scanner that uses a ring of thousands of transducers to capture vertical slices of muscle, fat, bone, and organs. A person steps onto a platform that lowers into a shallow pool, the sensors send ultrasonic waves through the body from every angle, and the reflected ripples are reconstructed into a detailed 3D image. The scan takes about 60 seconds. CEO David Holz said the system "aims for image quality comparable to MRI in many ways" and could be used as often as daily, and he described an eventual version "in many ways superior to even MRI machines" without radiation or powerful magnets. Those are the company's claims, and the gap between them and a cleared diagnostic device is wide.

The honest status is early. About a dozen people have been scanned. The scanner was built with ultrasound firm Butterfly Network, using 40 of its ultrasound-on-chip modules paired with two petaflops of processing. Holz wants to place ten units in a San Francisco spa opening before the end of 2027, alongside saunas, cold plunges, and a gym. Diagnostic applications will require FDA clearances the company does not yet hold; for now Midjourney says it is working on "body composition maps" that face a lower regulatory bar than diagnostic imaging.

The connection to Midjourney's image-generation work is loose, and the company hasn’t clearly defined that connection beyond suggesting it’s a secondary use for otherwise idle AI compute. That detail is the more revealing one. A company that built a large compute base to render pictures is pointing the same hardware at the inside of the human body, treating diagnostic imaging as an adjacent market it can enter laterally rather than a domain locked behind dedicated multimillion-dollar machines and the institutions that own them.

The capability here is demonstrated at a dozen scans, not deployed to patients, and preventive whole-body scanning carries real questions about false positives and incidental findings that the medical literature has documented for years. But the trajectory the effort points at is worth holding clearly. MRI-grade interior imaging has been scarce, expensive, and rationed through hospital schedules. An ultrasound rig that runs on commodity AI compute and aims for a 60-second scan at spa frequency describes a future where looking inside your own body shifts from a rare, costly event toward something abundant and routine. Whether Midjourney clears the regulatory path or not, the cost curve it is testing is the one worth watching.

The Other Side

For as long as there has been wealth, fixing its concentration has meant arguing over how to redivide a fixed amount of it. The Sanders bill is the most ambitious version of that argument yet: tax the largest AI firms by half, seed a $7 trillion fund, pay every American about $1,000 a year. The whole design assumes the pie is fixed and the only open question is who gets the slices.

The same people doing the arguing keep undercutting the assumption. Several of the figures fighting over how to divide AI's wealth, Musk among them, also say AI and robotic output will soon make that wealth far less scarce. Economists already estimate AI's measured output grew roughly 2,600% last year while the price of each unit of capability fell nearly as fast. Money is becoming a strange ruler for what is actually happening.

You can watch what it measures get cheaper by the day. The population living with brain implants doubled toward 150. A generative-image company pointed idle compute at MRI-grade body scans aiming for a 60-second cost. A developer paid to abandon wind put the money into geothermal because firm carbon-free power keeps getting buildable.

Imagine your own household in 2034. That $1,000 was meant to be a floor under real worry in 2026 - the scan you put off, the power bill you watched, the stressful mental math of whether you could afford for your kid to get sick. By 2034 the scan is a routine 60 seconds you barely schedule, the power on your block comes off something that no longer fouls the air, the diagnosis that took a specialist and a six-month wait arrives moments after you ask. The things the dividend was fighting to put within your reach got cheap enough that the fight over slicing reads strangely, the way arguing over who gets the bigger share of air would read to you now. That ease exists because in 2026 the wealth was being divided while it was already getting cheaper, wider, and harder to hold still. The slice was never the thing worth winning.

The Century Perspective

With a century of change unfolding in a decade, a single day looks like this: the population living with brain implants more than doubling toward 150 as an FDA-approved trial places its first speech-restoring device and a national approval clears the technology for use outside the lab, a generative-image company pointing its idle compute at a 60-second full-body ultrasound aiming for MRI-grade pictures on commodity hardware, a developer paid to abandon offshore wind routing the proceeds into next-generation geothermal because firm carbon-free power keeps getting buildable, a bill proposing to tax the largest AI firms by half and pay every American a yearly dividend, eighty Maryland lawmakers moving to push $1.6 billion in grid costs back onto the data centers that need the lines, three Amazon engineers taking their case for local buildout rules to a civil-rights office. There's also friction, and it's intense - an entire export-control apparatus straining to enforce a single machine's fence on an accusation its own officials will not show, a company opening an internal investigation into the engineers who testified for data-center transparency on their own time, $2.5 billion in public money spent to pull more than seven gigawatts of clean wind offline and steer much of it into gas, Klarna converting full-time service jobs into an on-demand pool of gig contracts, ratepayers across one state stuck with the bill for transmission serving another, a redistribution fight that would relocate AI's upside but only inside one country's borders. But friction generates texture, and texture is what finally gives a hand something to grip on a surface that kept sliding out of reach. Step back for a moment and you can see it: containment asserted from one capital through a single chokepoint - one company, one tool nobody else can build - while the capability it guards routes around the supplier entirely, and the upside, the cost, and the accountability of the buildout getting relocated toward the people living beneath it from underneath, one bill, one FERC complaint, one local statute at a time, as the medicine and imaging that genuinely widen human reach arrive cheaper, wider, and harder to hold still. Every transformation has a breaking point. Wind can rip the roof off the house... or fill the sails of everyone who learns to set them.

AI Releases & Advancements

New today

  • Perplexity: Launched Brain in research preview for Max and Enterprise Max subscribers on June 18, a self-improving memory system for the Computer agent platform that builds a context graph across sessions, files, projects, and decisions; Brain reviews the graph overnight to give agents full prior-work context at the start of each task, boosting answer correctness by 25% on repeated tasks and reducing context-heavy task costs by 13%. (Decrypt)
  • Model Context Protocol (MCP): Released Zero-Touch OAuth, an enterprise-managed authentication spec for MCP that lets organizations centrally provision, govern, and revoke AI agent access to tools without individual users managing OAuth flows; available now in the MCP specification. (MCP Blog)
  • TesterArmy (YC P26): Launched an agentic testing platform that deploys AI agents to run end-to-end automated checks on web and mobile applications before deployment and in production, available now at tester.army. (TesterArmy)

Other recent releases

  • Hugging Face / Microsoft / Google / GoDaddy (multi-org): Launched Agentic Resource Discovery (ARD), an open specification and reference implementation enabling AI agents to search for tools, skills, and other agents at runtime across federated registries; Hugging Face ships ARD support in the hf discover CLI, indexing thousands of Hub Spaces as discoverable MCP servers and AI skills via a POST /search REST API. (Hugging Face Blog)
  • Amazon Web Services: Open-sourced Strands Robots (Apache 2.0), a Python SDK that exposes the LeRobot robotics stack as AgentTools composable into a single Strands agent, unifying simulation, dataset recording, policy inference, and multi-robot fleet coordination through a shared interface runnable on a laptop without hardware or GPU. (Hugging Face Blog)
  • Allen Institute for AI (AI2): Released MolmoMotion, an open-weight language-guided 3D motion forecasting model built on Molmo 2 that predicts future 3D point trajectories of objects given a video frame and action description; available in autoregressive and flow-matching variants, with the accompanying 1.16M-video MolmoMotion-1M dataset and PointMotionBench evaluation benchmark. (Hugging Face Blog)
  • Framer: Released Framer 3.0 with Framer Agents, canvas-native AI collaborators that can design full pages, write code, manage CMS content, configure SEO, and adjust breakpoints directly within a team's existing Framer workspace; ships alongside Branching for non-destructive parallel experimentation and a new Community marketplace for creators. (Framer Blog)
  • Google: Released Android 17, repositioning Android as an "intelligence system" with Gemini Intelligence, an on-device AI agent capable of multi-step cross-app tasks such as converting a grocery list from Notes into a ready-to-purchase cart or pulling textbook requirements from Gmail; also ships Rambler AI voice-input filtering for Gboard and Gemini-powered custom home screen widget generation. (Android Developers Blog)
  • Adam CAD (YC W25): Open-sourced CADAM on GitHub, an AI-powered CAD tool for mechanical engineering that lets engineers create and iterate on designs through AI agents. (GitHub)
  • Lightricks: Released new updates to LTX-Video-Trainer on GitHub adding no-code custom AI video model fine-tuning, enabling creators and studios to train style-specific video models on proprietary footage locally without uploading content to external servers. (GitHub)
  • NVIDIA: Released the NVIDIA ACE Game Agent SDK in beta and a new suite of Unreal Engine 5 plugins at Unreal Fest 2026, enabling on-device AI companion development; the SDK is a C/C++ agentic framework for NPCs supporting Agent, Chat, and RAG APIs, while the UE5 plugins cover ASR (nemo-conformer-ctc-120m), SLM (Qwen 3.5 4B), and TTS (Chatterbox Turbo 350M) with Blueprint and C++ support. (NVIDIA Developer Blog)
  • NVIDIA: Released NVIDIA XR AI in public beta, an open-source library (github.com/NVIDIA/xr-ai) for building intelligent agents for AR glasses, XR headsets, and wearable devices; the framework connects XR devices to GPU-accelerated AI services via a modular media hub architecture supporting Cosmos visual grounding, Nemotron language models, and MCP enterprise tool connectivity. (NVIDIA Developer Blog)
  • Alibaba Qwen: Released Qwen-Robot Suite, three foundation models for embodied AI: Qwen-RobotManip (VLA model for manipulation built on Qwen3.5-4B VL, trained on 38,100+ hours of robot and human video data), Qwen-RobotWorld (video world model for predicting physical scene evolution), and Qwen-RobotNav (spatial navigation model); pilot testing underway with Alibaba Cloud enterprise clients. (Qwen Blog)
  • Wolfram Research: Released Wolfram Language and Mathematica Version 15 with built-in AI capabilities including a native AI assistant, symbolic music, and significant core functionality additions; now available to users. (Stephen Wolfram Blog)
  • NVIDIA Labs: Released cuTile Rust on GitHub, an open-source library enabling safe, data-race-free GPU kernel programming in Rust using NVIDIA's CuTe tile abstraction. (GitHub)

Sources and Further Reading

Artificial Intelligence & Technology's Reconstitution

Institutions & Power Realignment

Scientific & Medical Acceleration

Economics & Labor Transformation

Infrastructure & Engineering Transitions

The Century Report tracks structural shifts during the transition between eras. It is produced daily as a perceptual alignment tool - not prediction, not persuasion, just pattern recognition for people paying attention.

Read more