Set the budget
A public institution fixes the atmospheric operating range and the net carbon budget, and states the conditions under which that budget tightens.
AETHER / OPEN RESEARCH / NOAH HICKS
The atmosphere is shared. What if we managed it?
AETHER explores atmospheric carbon as public infrastructure: autonomous systems doing the work, useful industry inside a measured budget, and institutions accountable for the balance.
STATUS / CONDITIONAL WORKING PAPER
An ambitious research direction, with a smaller operating case you can inspect. No deployment or probability of success is claimed.
01 / START WITH ONE OPERATING SYSTEM
A planetary ambition.
A regional test of the idea.
1 Mt / yrIllustrative nameplate capacity
A regional direct-air-capture and storage benchmark. Actual output is limited by its resources and operating assumptions.
Same physicsTwo operating cases
Compare ordinary operations with assumed automation assistance. Electricity, heat and storage requirements do not disappear.
Open ledgerCarbon and cash, separately
See gross capture, lifecycle burdens, durability and funding. Change a constraint and the supported output must change with it.
This is an analytical example, not a proposed construction site or a validated plant design. The original 100 Gt/year global case remains a stress test, not the project’s predicted scale. Explore the paired case
02
THE PROPOSITION
Industrial carbon can be useful. Atmospheric capacity is finite. AETHER asks how a public system could measure, price, and settle the difference.
A capture machine provides one service. A working system also needs power and heat, storage, maintenance, a solvent operator and an independent account of what happened to the carbon. AETHER studies those connections through a regional example and a wider set of screening models.
The optimistic premise is that increasingly capable AI and robotics could make this infrastructure easier to build and operate. The research question is where that improvement survives contact with physical constraints, competing uses of resources and public consent.
03 / A PUBLIC CARBON UTILITY
A shared service.
A measured balance.
AETHER examines a public carbon utility: an institution that sets an operating budget, procures removal and carries long-term obligations. Public trusts, regulated operators and regional procurement arrangements have different authority and incentive problems. None is assumed to work merely because it is public.
Meter net use
Additions and removals are recorded by source, pathway, durability, location, and risk. A tonne nobody measured does not enter the ledger.
Price the load
Industry pays for permitted net use of a shared service. The price tracks scarcity, removal cost, storage durability, and system risk.
Maintain the system
Operating fees and separately identified legacy-removal funding cover service, monitoring and reserves. Independent verification keeps operators accountable.
Industrial activity remains possible.
Managing ongoing emissions and drawing down historical accumulation are different services. Their physical obligations and funding need to be accounted for separately.
04 / WHERE THE MODEL STRAINS
What determines
how far it can go.
These are the constraints the research has to connect. A correct calculation can expose a requirement; it cannot establish that the necessary infrastructure or institution will exist.
Energy
Engineering constraintPower + heatDifferent services, specified separatelyAnnual energy is only a first check. A real corridor needs hourly delivery, temperature-appropriate heat, grid access and additional low-carbon supply.
SourceStorage
Site evidence neededA real basinInjection rate is not the same as pore-space capacityWells, transport, permits, monitoring and long-lived liability have to fit a named site. No site-specific storage plan is established here.
SourceAutomation
Field evidence neededMeasured uptimeTask hours and reliability—not robot counts aloneThe regional comparison changes labor hours and availability explicitly. It does not assume that intelligence removes material or energy constraints.
SourceAccounting
Separate ledgersCapture ≠ creditKeep physical retention, emissions and risk separateTransport losses, project emissions, measurement uncertainty and reversal risk each reduce what the service can responsibly claim.
SourceFunding
Scenario assumptionTwo obligationsCurrent atmospheric use and historical cleanupA fee on current emissions does not automatically pay for legacy drawdown or post-closure care. Each obligation needs an identified funding source.
SourceClimate response
Projection withdrawnNo target dateAbsolute climate projections are quarantinedA failed baseline diagnostic prevents credible concentration and temperature claims. A historically consistent carbon-and-climate model is still required.
Source
AI + ROBOTICS
Coordination becomes physical only through machines, energy, materials, and operators who can be held to account.
05 / AUTONOMOUS INFRASTRUCTURE
Intelligence schedules the work. Something still has to do it.
AI could compress materials discovery, plant design, dispatch, anomaly detection, and system-wide optimization. Robots could build modules, service sites, inspect wells, maintain contactors, and gather field evidence.
The paired regional example asks a narrower question: what changes if maintenance takes fewer labor hours or equipment stays available longer? Those improvements are explicit assumptions, not demonstrated robot performance. Measuring them in the field is a useful next contribution.
AI systems
Search, model, allocate, schedule, monitor, diagnose.
Physical automation
Manufacture, construct, move, maintain, inspect.
Public institutions
Set targets, grant permission, assign liability, stop the system.
06 / THE ATMOSPHERIC NORTH STAR
From 428.73 ppm toward roughly 280.
NOAA’s preliminary May 2026 global monthly mean is a dated observation, not a live feed. Roughly 280 ppm is our long-horizon restoration aspiration, not a demonstrated safe or optimal modern setpoint. Getting there is a carbon-cycle and governance problem, not subtraction.
MAY 2026 / NOAA GLOBAL MEAN428.73 ppm
0.042873% of dry air · 1.53× the 280 ppm anchor · about 53% above it
LONG-HORIZON RESTORATION NORTH STAR≈280 ppm
0.028% of dry air · a 148.73 ppm gap from the dated observation · about 34.7% below that concentration
Open question
No credible arrival date is established here. Absolute climate projections have been withdrawn after a failed baseline diagnostic. A responsible trajectory must account for land, oceans, other climate forcings and ecological effects.
NOAA measurement source07 / CARBON AFTER RESTORATION
The target is a control boundary, not a finish line to overshoot.
As concentration approaches an agreed operating range, the removal system should throttle down. Captured carbon can become feedstock for durable materials or tightly closed industrial cycles, provided lifecycle accounting shows where it went and for how long.
Carbon use is not automatically durable removal. Fuels and short-lived products can release it again; long-lived products need a verified fate. Product revenue also cannot be assumed to fund the entire removal system.
08 / WHAT AETHER MEANS
The name focuses on the details.
Managing atmospheric carbon at scale takes physical infrastructure, public authority, and an enormous amount of clean energy. The acronym says so out loud.
A
Atmospheric
The medium whose carbon concentration the system manages.
E
Engineering
A maintainable capability, measured — not an offset promise.
T
Through
The link across energy, industry, storage, markets, and law.
H
High-
The real industrial burden, even in an automation-rich future.
E
Energy
The input that intelligence cannot remove from the equation.
R
Removal
Verified drawdown, used to keep the budget balanced over time.
AUTHORITYWho sets the target?
Public institutions, affected communities, and independent science. Not an autonomous fleet.
EVIDENCEWhat counts as removal?
Net, additional, durable, independently measured carbon, with lifecycle emissions and reversal risk included.
STOP CONDITIONWhat ends an intervention?
Atmospheric thresholds, ecological harm, failed verification, revoked permission, or a safer alternative.
09 / AI SCENARIO CONTEXT
Borrow the timelines. Don’t borrow the certainty.
These projects explore different AI futures. They provide scenario context, not measured inputs for construction speed, robot reliability, capture energy or storage throughput.
Fast capability branch
Situational Awareness
Stress-tests rapid cognitive progress, datacenter power demand, and industrial mobilization.
Short-timeline branchAI 2027
Asks how little time institutions and physical supply chains might get.
Governance-bounded branchAI 2040: Plan A
Tests slower scaling with transparency, verification, and coordinated control.
10 / THE LONG VIEW
A low form of terraforming, starting with responsibility on Earth.
Deliberately changing an atmosphere at scale is a restrained form of terraforming, and that framing raises the burden of evidence and consent rather than lowering it. Descendant systems for autonomous construction, resource cycling, atmospheric processing, and remote verification might one day matter on other bodies. Earth is where the governance has to be learned first.
11 / OPEN THE MODEL
Help turn the idea
into evidence.
Start with a specific question: check the regional energy balance, replace an automation assumption with field evidence, or test the utility’s funding under failure. The working paper is not externally peer reviewed. The review guide identifies contributions that could change its conclusions.
What changed in v0.46