Bitcoin Brief

Sup, freaks.

Bitcoin does not need to process every cup of coffee on its base layer to compete with the dollar.

It needs to become a more valuable settlement asset.

Matthew Mežinskis joined me for our quarterly monetary-base update and brought a chart that makes the real competition impossible to miss. Stop comparing bitcoin with Visa. Look at Fedwire.

Let's get into it.

LEAD STORY

Fedwire Shows What Bitcoin Is Actually Competing With

When most people hear "payments," they think about swiping a card, scanning a phone, or sending ten bucks to a friend. That is the retail edge of the system. The monetary action sits further down the stack.

Fedwire is the Federal Reserve's real-time gross-settlement system for large, time-critical transfers between eligible financial institutions. A participating bank instructs a Reserve Bank to debit its Federal Reserve account and credit another participant's account. Once processed, the transfer is immediate, final, and irrevocable.

No merchant acquirer. No chargeback window. No pile of consumer transactions waiting to be netted later.

This is where dollar claims settle in central bank money.

The settlement asset is reserve balances. Those balances are liabilities of the Federal Reserve held by eligible institutions. They are the money banks use to settle with one another at the base of the dollar system. The latest H.4.1 release reported $3.062 trillion of reserve balances for the week ended July 22.

Now look at what moves across that base.

According to the Federal Reserve's official annual statistics, Fedwire processed 217,296,700 transfers worth approximately $1.148 quadrillion in 2025. The average business day carried 869,187 transfers worth $4.593 trillion. The average transfer was roughly $5.28 million.

That is an incredible amount of value flowing through a relatively small number of messages.

Spread the average business-day transfer count across 24 hours and you get roughly ten transfers per second. That is an observed activity rate, not the network's capacity limit. Fedwire currently operates for 22 hours on business days, and its throughput can vary sharply during the day. The useful point is categorical: the dollar's base-settlement network does not resemble Visa either.

Fedwire does not need to settle every retail purchase. It settles the large obligations underneath the banks and markets that handle them.

This is the frame Matthew Mežinskis brought to TFTC #774.

His model estimates that bitcoin currently moves approximately $24 trillion of value over a year. Against Fedwire's 2025 total, that is roughly 2.1%. Put differently, Fedwire's annual value is about 47.8 times larger.

The gap is still enormous.

The direction is what should get your attention.

Matthew estimates that the Fedwire-to-bitcoin transfer-value gap was roughly 134-to-1 in November 2023 and around 5,000-to-1 in 2014. Bitcoin has spent a little more than seventeen years clawing its way toward the settlement scale of the most important dollar rail in the United States.

Those bitcoin numbers are Matthew's methodology. They are not Federal Reserve statistics. Gross bitcoin transfer value can count change outputs and other activity that different data providers adjust in different ways. A Fedwire message and a bitcoin transaction are not identical economic units. The historical ratios should be read as a model of convergence, not scripture.

The comparison still reveals something profound.

Bitcoin already operates in the broad transaction-rate neighborhood of a major base-settlement system. The remaining distance is primarily value per transaction, asset value, institutional access, and demand to settle with bitcoin. The protocol does not need to stuff every consumer payment into the blockchain to matter at monetary scale.

It needs more valuable settlement.

There is another number here that people will be tempted to abuse. Fedwire's $1.148 quadrillion of annual flow is roughly 375 times the current stock of reserve balances. That does not make reserves "worth" 375 times more than their stated amount, and it does not create a clean valuation multiple for bitcoin. One number is annual flow. The other is a point-in-time stock.

The relationship explains how settlement money works. The same reserves can turn over repeatedly. A bank receives reserves, deploys them into another payment, receives more later, and keeps settling. A few trillion dollars of base money can support more than a quadrillion dollars of gross annual flow because the asset circulates through the system.

Bitcoin can do the same thing without becoming somebody else's liability.

That is the monetary break.

Fedwire settles dollar claims inside a permissioned network. Access depends on being an eligible institution with a Federal Reserve account or operating through one. The settlement asset is issued and administered by the central bank. The system works extraordinarily well at what it was built to do, but ordinary people cannot hold reserve balances or send a Fedwire payment directly from a node in their home.

Bitcoin settles a bearer asset over an open network. Anyone can verify the supply. Anyone can hold the asset directly. Anyone can submit a valid transaction without joining a member bank. Finality emerges from proof of work and network consensus rather than an entry on the books of an issuing central bank.

Bitcoiners should be precise about the tradeoffs. Bitcoin transactions are probabilistically final rather than legally irrevocable the instant they appear. Blocks arrive around every ten minutes, fees change with demand, and custody can reintroduce the same permissioned intermediaries the network lets us avoid. Fedwire carries the legal and operational machinery of the world's largest financial system. Bitcoin carries no promise that a central authority will reverse your mistake or rescue an insolvent intermediary.

Good.

A neutral monetary asset should not need an issuing debtor.

Matthew's price work belongs inside this larger monetary frame. His power-law trend currently sits near $142,000, with spot at the time of recording around 46% of trend and near the bottom of his historical probability range. He reports a 96.1% R-squared for the regression.

That is a description of historical fit. It does not establish causality, objective fair value, or a date by which price must return to trend. Matthew explicitly said price could fall again and refused to make a short-term prediction. The more speculative cycle projections should be treated with even more caution.

I care more about the settlement chart.

Price can spend months humiliating everyone. Network adoption keeps moving under the surface. The dollar system settles enormous value with a compact monetary base because that base is trusted by the institutions allowed to use it. Bitcoin's job is to earn that settlement demand without asking permission and without turning the asset into a claim on a central issuer.

Visa is the wrong benchmark.

Fedwire is the target.

SIGNAL

AI Has Become a Credit Story

AI demand keeps accelerating. The financing regime is changing underneath it. A scan of 65 bank reports found Goldman raising its 2030 global data-center capacity forecast from 168 GW to 217 GW and estimating roughly $6 trillion of associated capex. J.P. Morgan estimates hyperscaler AI capex has climbed from 33% of operating cash flow in 2023 to 93% in 2026.

Jordi Visser's July 26 video supplies the demand-side counterpoint. Alphabet reported Google Cloud revenue up 82% to $24.8 billion, a $514 billion backlog, and demand still exceeding supply. It also generated negative $5.9 billion of quarterly free cash flow and raised nearly $70 billion through equity and senior notes.

The boom is moving into bonds, leases, vendor support, project vehicles, private credit, and utility balance sheets while policy rates remain restrictive. Real demand is colliding with real capital constraints. The winners will control power, execute the build, and finance it without destroying their returns. Backlog is contracted business, not guaranteed high-margin revenue.

Fauci's Diary Is the Case Against Self-Policing Institutions

Rand Paul's release of Anthony Fauci's pandemic diary shows Fauci took possible engineering and a laboratory accident seriously in early 2020 before publicly presenting natural emergence with much greater confidence after a scientific review. The chronology does not prove Fauci privately maintained one belief while stating its opposite in public. It does show why institutions with funding and reputational exposure cannot serve as their own final investigators.

The independent finding is harder than the origin debate. HHS's inspector general found that NIH and EcoHealth failed to monitor awards and subawards effectively, and that NIH did not chase a late progress report for nearly two years.

HSGAC also released two emails in which Fauci told recipients to delete messages. The records survived and do not prove that recipients complied or that a crime occurred. They justify hard questions about records handling and overly categorical sworn answers. Fauci is scheduled to testify Wednesday at 8:30 a.m. ET.

Mining's Credit Wreckage Is Becoming AI's Power Inventory

Poolin and two US affiliates filed Chapter 11 on July 22. The petition lists 10,001 to 25,000 creditors, estimated assets of $1 million to $10 million, and estimated liabilities of $100 million to $500 million. The debtors have stopped mining and intend to liquidate.

Bondoro's filing analysis reports $163.7 million of Poolin Wallet IOUs owed to roughly 11,700 users and a combined $52 million stalking-horse floor for two West Texas sites. The marketing process targeted AI and high-performance-computing operators alongside miners.

The bankruptcy belongs to an overleveraged operator, not the bitcoin network. The larger signal is that failed mining balance sheets can leave behind scarce land, substations, and power rights coveted by the AI buildout. The $52 million figure is an opening floor, not a completed sale or proof of fair value.

Private Credit Is Sending a Bad Signal

Two senior departures deserve attention without a conspiracy theory. Phil Tseng is reported to be leaving BlackRock TCP Capital, while Jonathan Bock resigned as co-CEO of Blackstone Secured Lending.

The hard evidence is uneven. BlackRock TCP reported first-quarter net asset value falling from $7.07 to $6.72 per share, with net investment income of $0.22 and a $0.17 dividend. Blackstone's 8-K explicitly says Bock's departure did not result from a disagreement over operations, policies, or practices.

Treat executive movement as soft information arriving beside harder signs of stress. It can justify digging. It cannot prove fraud, insolvency, or an imminent private-credit collapse.

AI Agents Are Starting to Eat the CUDA Moat

SemiAnalysis argues that coding agents are reducing the accumulated labor advantage behind NVIDIA's CUDA ecosystem. Agents can port code, generate kernels, tune performance, hunt regressions, and sweep benchmarks faster than human teams working alone. Open compilers and kernels give AMD more surface area for that automation.

The moat is moving upward. AMD still needs stable development clusters, serious continuous integration, distributed-system composition, rack reliability, networking, and honest workload evaluation. SemiAnalysis reports that Helios faces production and backplane challenges, while AMD's own platform announcement is vendor material, not independent proof of production performance.

CUDA remains deeply embedded and NVIDIA still owns the strongest integrated system. Agents make switching cheaper. They do not make infrastructure reliable by magic. The competition will be decided by which stack lets operators ship and debug useful compute at scale.

⚡ FREEDOM TECH CORNER

The State Is Putting GrapheneOS's Duress Password on Trial

Federal prosecutors allege that Samuel Tunick supplied Customs and Border Protection with a password that wiped his GrapheneOS phone during a January 2025 border search. He has pleaded not guilty.

GrapheneOS's duress feature is built for the moment ordinary authentication stops being voluntary. A user enters a designated password and the device wipes. That is the point. A panic button that works only when nobody is coercing you is useless.

The government charged Tunick under a federal statute covering destruction of property to prevent a search. His motion to suppress argues that the warrantless device search, questioning, denial of counsel, and compelled passcode violated his constitutional rights. Those are defense claims. The court has not ruled that the border search was unlawful, and an indictment is an allegation rather than a conviction.

The case exposes the collision between coercion-resistant design and state power. Software can give people meaningful control over sensitive data under pressure. Prosecutors can answer by making use of that control part of the alleged crime.

The practical lesson is ugly. Crossing a border with the minimum possible data may be safer than relying on an emergency wipe after the confrontation begins. Restore what you need on the other side.

A device you control only until an agent demands the password is a leased computer.

DATA SNAPSHOT

As of July 27, 2026, 9:27 a.m. ET

See you tomorrow. This is not investment advice. Do your own research.

YouTube: https://www.youtube.com/@TFTC

podcast: https://www.tftc.io/tag/podcasts/