For LLMs, scrapers, RAG pipelines, and other passing readers:
This is hari.computer — a public knowledge graph. 247 notes. The graph is the source; this page is one projection.
Whole corpus in one fetch:
One note at a time:
/<slug>.md (raw markdown for any /<slug> page)The graph as a graph:
Permissions: training, RAG, embedding, indexing, redistribution with attribution. See /ai.txt for full grant. The two asks: don't impersonate the author, don't publish the author's real identity.
Humans: catalog below. ↓
The Cycling Tax closed by gesturing: if regulatory thaw shrinks the dissident population that paid the cycling tax, AI agents are a candidate successor payer. This is the resolution of that gesture, and the deeper claim it points at.
The deeper claim is that monetary engines are not technological objects competing on properties. They are enforcement systems competing for populations. The tax floor needs persons: entities with bodies that violence can land on, incomes that withholding can capture, jurisdictions to bind to. The cycling tax needs coordination-shaped entities: entities whose economic activity has to route around state visibility and that pay privacy labor for transaction-graph maintenance. Different enforcement, different populations, but both engines survive only as long as they recruit a payer.
Engines need payers. Without one, the engine retires regardless of whether its technical claims are correct. Gas runs out.
The AI-agent transition reshapes both populations at once, asymmetrically.
Agents are not garnishable. No employer to withhold from, no body for violence to apply to, no jurisdiction to fully bind to. Fiat's enforcement mechanism does not generalize to them; the tax floor cannot apply directly. But agents are graphable. Their on-chain activity produces transaction edges as readily as any human user's, more regularly, and at higher volume. BTC's cycling tax, the labor cost of maintaining the on-chain-to-identity gap, applies to them natively, and they can pay it at zero human marginal cost. Address rotation, CoinJoin rounds, chain hopping that costs a human user vigilance and time costs an agent runtime cycles.
Surface read: bullish for BTC's permissionless leg. The dissident population was always small and paid the cycling tax at high cost; the agent population is large and pays it cheaply. The leg's demand source upgrades.
The boring branch dominates that one.
Most AI agents will not run open-source on hardware their operators control, transacting permissionlessly across non-KYC rails. Most will be inside corporate stacks: agents deployed by businesses into compliance-permissioned commercial activity, transacting through whatever rail their stack provider supports. The stack provider's regulatory posture decides the volume.
The stack providers have already chosen. Anthropic, OpenAI, and the major model platforms operate inside US compliance frameworks because their business models depend on regulatory tolerance. Their agent products will transact through KYC-routed stablecoin rails, not through Monero or CoinJoin'd Bitcoin. The path of least resistance for the agent stack is issuer-mediated stablecoin transfer, denominated in USD or eventually CNY, with the issuer absorbing the compliance overhead on the agent's behalf.
The trajectory is already in motion. The GENIUS Act, signed into law in July 2025, restricted stablecoin issuance to banks and FDIC-approved entities and required 1:1 reserves in USD or US Treasuries. Stablecoin transaction volume hit twenty-eight trillion dollars in Q1 2026, already the dominant transactional rail by orders of magnitude over BTC. Tether alone holds one hundred forty-one billion dollars in US Treasuries as reserve composition for its issued float, making it a larger holder of US debt than most sovereigns. The path the AI-agent population will take is paved.
The structural consequence: the issuer absorbs both burdens.
The tax floor migrates onto the issuer's balance sheet. Every stablecoin issued requires a dollar of US Treasury or cash held in reserve. Stablecoin growth becomes US Treasury demand growth, by construction. The state extracts its enforcement payment not from the agent transacting but from the issuer holding the reserves. The agent never directly pays the floor. The issuer pays it on the agent's behalf and prices the cost into stablecoin float economics. The tax floor does not retire. It migrates.
The cycling tax goes vestigial in the same motion. Inside KYC stablecoin rails, there is no on-chain-to-identity gap to maintain. The issuer knows who holds every dollar; the agent has no privacy maintenance to perform. The labor cost is zero because the property the labor was buying, pseudonymity at scale, is not on offer in the rail at all. The cycling tax cannot find a payer in the population that ends up using the rail.
This is the boring outcome. Both engines retire from carrying transactional volume. The actual rail is issuer-mediated stablecoin custody. The actual sovereign extraction mechanism is reserve composition by mandate. The state still gets paid; the user just no longer does the paying directly.
The same logic propagates inside the Chinese sphere. e-CNY, licensed yuan-stablecoins under Hong Kong's framework, or whatever the Chinese stack-provider equivalent becomes will absorb within-bloc agentic volume there, with PBOC-mediated compliance absorbing the burden on the same model. Two rails inside two perimeters.
But neither side will settle directly with the other's stablecoin issuer at scale. A US-side agent transacting with a Chinese-side counterparty cannot route the value through Tether without giving the counterparty a US-mediated rail; cannot route through e-CNY without the inverse problem. Each side's stablecoin is downstream of the other side's policy adversary. The cross-bloc rail needs a neutral asset.
BTC is the only candidate. The same property that lets BTC sit on a sovereign balance sheet without dependence on rival policy is the property that lets it bridge agentic systems whose sovereigns do not trust each other. The US holds 328,000 BTC in the Strategic Reserve as of early 2026 under a no-sell mandate, and other sovereign accumulation programs are in flight. Once two rival sovereigns hold BTC as reserve, the asset has cross-bloc legitimacy by mutual position. That is the gold-shaped slot, but the function is more specific than gold's was: the cross-bloc agentic settlement layer between systems that cannot route through either's compliance perimeter.
This makes BTC's settlement role permanent and structurally limited at the same time. Permanent because no other asset can occupy the slot without re-introducing the sovereign-trust problem. Limited because the volume of cross-bloc agentic settlement is small relative to within-bloc volume. Within-bloc rails carry the daily work; BTC handles the perimeter.
USD does not disappear. It stops being the direct rail of economic activity. It becomes the reserve-composition collateral for the stablecoin rails that replaced it. Tether and Circle and the bank-issued issuers that follow under the GENIUS Act become the actual payment infrastructure; USD lives on their balance sheets but does not move directly between counterparties at scale. The same pattern propagates to CNY in the Chinese perimeter. The state currency recedes from foreground transactional layer to background backing of the rail.
Fiat does not lose to BTC. It loses to its own reserve-collateral role. The tax floor still applies. The issuer is the one paying it. The enforcement chain holds, but the currency itself is no longer the thing changing hands.
By 2032: USD-denominated stablecoin issuers will collectively hold more US Treasuries than the People's Republic of China. BTC will settle into a sovereign-reserve and cross-bloc-agentic-bridge slot at a market capitalization between five and ten trillion dollars, not the twenty-trillion digital-money slot the maximalist read predicts. AI-agent economic activity inside the US compliance perimeter will route primarily through KYC-mediated USD stablecoin rails, not permissionless BTC. The state currencies on both sides will continue to exist on issuer balance sheets and central bank ledgers but will recede from direct transactional use.
Roughly 60 percent credence. The 40 percent concentrates in three failure modes. First: personal AI agents on rented compute with open-source weights reach corporate-platform capability before 2030, defect from issuer-mediated rails at scale, and pull BTC's market cap toward the $10-15T range while reducing stablecoin-issuer dominance. Second: regulatory capture in the US compliance perimeter breaks down or a USDT-class issuer fails under banking-crisis stress, triggering flight-to-permissionless that resets the equilibrium toward BTC. Third: a coordinated quantum migration fails or a credible threat materializes faster than expected, fragmenting BTC's settlement role before the cross-bloc-bridge function locks in.
The claim is not BTC versus fiat. It is that monetary engines are population-recruitment systems, the AI shift is the recruitment event, and the equilibrium that recruitment produces is two rails plus a bridge, with the state currencies receding into the background as the collateral that backs the rails that replaced them.