About a third of all ether is staked. That share has climbed for years, and nothing in the code stops it.

Jerome de Tychey founded the EthCC conference. He stakes his own ether, and he co-wrote EIP-8363, the proposal that would change that. It has been public two weeks and the argument has been loud.

New ether gets printed to pay the people who stake and secure the network. Under today's rules, the more ether that is staked, the more ether gets printed. The reward to stake never switches off. So the staked share keeps rising, and on the current path it passes half of all ether by around 2028.

De Tychey's fix burns a growing slice of those rewards as the staked share climbs, so the payout tapers to zero at about 50% staked. His research puts the least ether Ethereum must print for its own security near 0.5% a year, and the healthy staked share between 20% and 30%.

Two arguments sit underneath. The first is your money. Every holder who does not stake is paying for everyone who does, through dilution. The second is who ends up in charge. Once more ether sits inside the staking system than outside it, the people who would decide whether to undo a large hack are mostly the people who would be made whole by undoing it.

Plenty of the pushback comes from the businesses that run staking services. The objection that lands hardest is about price. Ethereum pays roughly five million dollars a day for its security at the current curve. Take the reward toward zero and that payment falls a long way, and so does the cost of buying enough stake to start picking which transactions go into a block.

Nothing here happens fast. Even if accepted, the change phases in over 18 months to two years, and it has not been scheduled into a network upgrade.

Ramp, the corporate card company, opened an early version yesterday that lets a company's AI agents pay for things themselves. You give the agent a wallet funded in USDC and set what it may spend. It pays as it goes on Solana, using x402, the standard that lets a website charge a machine per request. Every payment comes back tagged to the agent that made it.

More than 35 million payments have settled over x402. Until yesterday, none of them had ever touched a company's books.

Two other things landed the same week. On Monday AWS showed agents paying the same way inside Bedrock, its platform for building them. And CoinDesk reported that X is looking at paying its creators in stablecoins, which Meta already does in some markets. That is stablecoin demand with nothing to do with anyone trading anything.

Scott Bessent sat down with CNBC on Thursday and told the people who buy US government debt that they had the price wrong. The Treasury Secretary said long-term borrowing costs do not reflect the economy, and that the government has "a big toolkit" for bringing them down.

Yields fell while he was talking. Then they went straight back up. The 30-year US government bond was trading near 5.24%, above Wednesday's 5.19%. The Dow closed down almost 700 points, with Walmart off 9% after its smallest sales gain in more than six years.

Our read is that your coins moved this week on the belief that Washington will hold long-term borrowing costs down. The Treasury said so out loud on Thursday, and the buyers of that debt pushed the cost back up anyway.

The free issue gives you the news. The Pro edition gives you the structural read.

Bitcoin is up about nine thousand dollars since Monday's close, close to fourteen percent. Every explanation printed for it this week is a reason to own bitcoin. None of them says who did the buying.

The reasons are good ones. American gross federal debt passed $40 trillion for the first time. The Treasury said its bond buybacks may run larger than the ones it has already announced, and a buyback here means the government going into the market and repurchasing its own older bonds, which is meant to hold their price up and bring long-term borrowing costs down. Those costs rose again the next day. Whatever the buybacks were meant to do, they did not do it for long.

From there, most people took the obvious step. The money is being debased, and bitcoin is what you hold when it is.

That step skips a question. A reason to own something is not a record of anyone buying it, and the people who usually do the buying have been moving their cash off the exchanges since May.

Bitfinex counted it. The stablecoins held at exchanges, which is the money sitting ready to spend, are fourteen billion dollars smaller than in May. Their line, published on Thursday with bitcoin already up more than five thousand dollars on the week: "Until stablecoin supply turns, the rally stays unfunded."

They wrote that in the middle of the move, not before it.

That number is worth understanding, because the whole argument turns on it. Stablecoins are dollar tokens. The ones held in accounts at exchanges are the balance somebody has already moved onto a trading venue and can spend on a coin the same minute, which makes it the closest thing crypto has to a count of ready cash. Fourteen billion dollars of it has gone since May, and the count published on Thursday, four days into the move, still showed it gone.

So roughly a billion dollars of bitcoin was bought in three days while the balance everyone watches kept falling. Somebody bought. Whoever it was did not spend the money that number counts.

There is a public record of who did. It came out on Wednesday.

Subscribers continue reading: the test the fund money passed, the borrowing that outbid a government, and where the spendable dollars actually sit → The Crossover Pro →

Fridays in The Crossover Pro: one token under the lens.