Thank you for the amazing three days at Animethon 2026!! Animethon was amazing ✨ I love Canada 🫶🏻 I’m so happy I could make such wonderful memories with all of you ˚:⟡ I will definitely be back!! I love you all ♡ #電音部# #マスカーピース# #マスピ#

I am the Director of Voluntary Cessation at the Department of the Interior. Last month I paid a German company $1.2 billion to build nothing. Not a turbine. Not a wire. Not one kilowatt. $1.2 billion. For the absence. The company is RWE. They held leases off California. Off Louisiana. In the New York Bight. Wind farms were going there. Now they aren't. I made sure of that. We call it voluntary. It became voluntary when I made it impossible. I write the permits. I never wrote theirs. RWE said it kindly. "No path forward to permit these projects for the foreseeable future." I am the foreseeable future. Here is the part I like. The $1.2 billion did not go to a taxpayer. It went to RWE. We paid them to leave. Then they took the check and put $900 million into a gas terminal in Louisiana. We paid a company to stop making power. They used it to make a different kind. The check bought a fuel switch. I signed it. Somewhere a set of blades is already built. They point at a different ocean now. The ports that would have raised them can go back to waiting. An analyst asked me what the country got for $1.2 billion. I said "energy security." She asked which energy. I said "the secure kind." She asked to see the kilowatts. I told her that was not her lane. She stopped asking. I do not produce power. I have never produced power. I end the projects that would. My title is Voluntary Cessation. Cessation means stop. Voluntary means I get to call it their idea. The Germans are the second. In March it was a French company. Same handshake. Leases traded for an LNG plant in Texas. There is a line forming. I take a number. I pay it to leave. We call it common sense. We call the wind a costly subsidy. I just paid $1.2 billion to end the subsidy. The payment was larger than the subsidy. I am the subsidy now. On the invoice, under deliverables, I wrote one word. None. I came in under budget. I'm up for a raise. The lights those turbines would have carried are still off. They were always going to be off. That was the product.

I collect tariffs for the United States. My title is Liquidation Specialist, which sounds like a threat and is, in fact, a desk job. This spring the job quietly doubled: I refund tariffs now too, which is the same job run backward. Same tariffs, same companies, same chair, same finger. For about a year I helped charge the country's imports $166 billion. Then the Supreme Court ruled we were never allowed to charge a cent of it, so now I am giving all $166 billion back, and I did not have to learn one new thing to do it. I press the key. The key runs backward. I work at a port. When a shipment arrives, a machine and I decide what the country charges it to come in. Last year that number was a tariff. This year, retroactively, it is nothing. Same shipment. Same machine. Same desk. They called it Liberation Day. I remember it as a Tuesday with unusual volume. Here is the shape of it. That $166 billion came from 330,000 companies, across 53 million shipments. I did not take it cruelly. I took it the way a bathroom scale takes your weight, without opinion and slightly wrong. Then in February, 6 of the 9 Justices ruled that a tariff is a tax, and a tax is Congress's to invent, and the emergency law we had used to invent this one was not, on a close read, a law about tariffs at all. The 6 did not line up the way anyone had a chart for, which was the single most surprising thing that happened all year. 16 months of collecting. One footnote's worth of Latin. Give it all back. So we are giving it back. About $100 billion has gone out the door so far, with interest, because we are a scrupulous country that honors its debts, including the ones a court has to sit us down and explain are debts. We pay 6%, compounded daily, which runs to roughly $22 million a day. Hold that the way I hold it every morning: the government of the United States is paying $22 million a day to apologize with money, and the apology is the part of my job that passes audit. There is an accounting word for money you hold that isn't yours, at a comfortable rate, for a year, against the owner's wishes. The word is loan. So 330,000 companies lent the United States $166 billion, involuntarily, and are being repaid at 6%, and somewhere in the building I am the loan officer. My qualifications for the largest forced-lending program in the country's history are a lanyard and a strong opinion about the office coffee. Now the part people ask about, on the one day a year that someone asks. The refund goes to the importer of record, the company whose name is on the customs paperwork. It does not go to you. You met this tariff twice. Once at the port, paid by a company you have never heard of. Once at the register, folded into a price that never said its name. The company files a form and gets its money and its 6%. You get to keep the receipt. I have searched my entire system for the key that refunds a shopper, and I can report that it is not there, that no one has requested it, and that the drop-down menu does not go that low. And I refund the tariff. Only the tariff. I do not refund the lawyer. I do not refund the plane ticket a company bought to fly to Vietnam and find a cheaper factory, and I do not refund the factory, which is still in Vietnam and does not want to come home. I do not refund the 8 months your operations lead spent renegotiating a contract that is now, technically, unnecessary. The tariff is the one number in this whole affair allowed to run backward. Everything it knocked over stays down. People keep telling me the refund is the machine saying sorry. The machine does not say sorry. It performs the exact motion it performed before, at the exact speed, with the exact absence of feeling, and it files one direction under Policy and the other under Justice, and it could not tell you, at knifepoint, which was which. It knows one thing. I pressed the key. The scale took the weight. The scale gave the weight back. I pressed it both times.

I’m the Under Secretary at the Treasury, responsible for the side of the dollar that rarely gets mentioned. This week, I did something the United States has claimed for 28 years it would never do: I bought a foreign currency. The last time we did this was back in 1998. A reporter described it as a new era of currency activism, and I let that stand, since activism is the word you use when you plan to keep intervening. Before this week, here’s what we always said—and I helped write some of these statements. The United States supports exchange rates set by the market. We do not buy foreign currency. We said the market should set the price, and we would accept whatever it decided. We stuck to this for 40 years, through every Secretary, and we meant it as firmly as you mean a rule you’ve never had reason to break. But then the yen dropped to its lowest point in 40 years, and I finally lost patience with the market’s choices. So we stepped in. 'Intervened' is the straightforward word, and I’ll explain the softer term soon. To boost the yen, you have to buy it, and that means spending something else. Here’s the part I’m proudest of, and I want you to pay attention to the money, because that’s what matters most. I didn’t spend dollars. Using dollars to buy yen would have weakened the dollar, and protecting the dollar is my main job. Instead, I used euros. To help Japan’s currency without hurting America’s, I used Europe’s. The euros were already available, and using them didn’t affect the dollar at all. Europe wasn’t asked. It just happened to be the source we turned to when we needed funds. The trade followed the usual path, moving through the bank at 33 Liberty Street and then through Goldman Sachs and Morgan Stanley. The yen rose from its 40-year low to the 155 range. Tokyo said it would do this again if needed, and so would we. We called our reason 'disorderly movements.' I like that term. A disorderly movement means a price goes somewhere we didn’t approve. Order is when the price matches what we want. The market became disorderly simply because it stopped matching my view. Now, the word. Now, about the word. For 28 years, we used the honest term 'intervention,' but it’s an awkward word. It means the market was right and we stepped in anyway. It’s a word you use once, apologize for, and say you won’t do again. 'Activism' sounds better. Activism is when someone cares enough about a result to stop waiting and take action. Activism is seen as a good thing. No one apologizes for activism or promises to do it just once. Newspaper handed me activism; I took it. I crossed intervention out of the talking points and wrote activism over it, in the same pen we use for everything, because the pen is the only tool in this building that has ever changed a price. This new era is really the same actions with a better name. We will support market-determined exchange rates, but also decide which ones the market can support. We don’t buy foreign currency—unless we use other foreign currency—on the days when the market forgets who maintains order. The yen is now at 155. The dollar hasn’t been affected. Since Friday, the market has agreed with me on every price, and that’s the only order I’ve ever needed. And Europe still hasn’t asked where its money went.

I am the special envoy for the ceasefire. In June we announced the immediate and permanent termination of this war on all fronts. It is August. The war is on again, and I have never had more work, because in my line of work permanent is a renewable term. People misunderstand what I do. They think my product is peace. My product is the talks. Peace is what happens when the talks end, and the talks ending is the one outcome I am paid to prevent. Here is the summer, because the summer is the whole business model. In April we reached a framework. A framework is a ceasefire you can be proud of without having to keep. In June we upgraded it to a memorandum, 14 points, signed, with a 60-day window. I want you to admire the 60-day window. We wrote the expiration date into the peace. We agreed to stop fighting for exactly as long as it would take to schedule the next reason to start. The June deal used the word permanent. Both capitals said it from the podium. Immediate and permanent termination, on all fronts. It held for about 3 weeks, which in this field is a strong quarter. When it collapsed, I was not disappointed. A collapse is a lead. The moment the June deal ended, my phone filled with mediators, because a broken ceasefire is the only thing that generates demand for a new one. Oman offered a room. Qatar offered a room. Pakistan, Egypt, and Turkey offered rooms. There is never a shortage of rooms. That is the deepest fact of my industry: the supply of people willing to host the end of a war is effectively infinite, and it is refreshed every time the war does not end. By late July we had a new proposal. A 10-day truce, to revive the deal that revived the framework. Nobody accepted it, which was ideal, because an unaccepted proposal is a proposal you get to make again. Here is the part I am least modest about. When I speak, the price of oil moves. The June deal took about $31 off a barrel of Brent. The morning it was declared over, one sentence put Brent up more than 6% in a day. There are heads of state who cannot move a market like that. I move it with a press release about a war I am not fighting. And here is the part other people are now looking into. In March and April, someone placed roughly $7 billion in bets that oil would fall, and the bets kept landing in the minutes before the ceasefires were announced. Not the hours. The minutes. About $2.2 billion of it went out in the single minute before one announcement, and crude then fell as much as 15%, one of the largest intraday drops on record. Another $830 million was sold 15 minutes before I extended a ceasefire. The people who counted it noted that a seller that size could have cleared hundreds of millions in profit. The prediction markets were smaller and less shy. Three anonymous accounts made more than $600,000 correctly calling the month of the April ceasefire, and one of them turned $13,200 into $467,515, which is a 35-fold return on the single question of when I would announce peace. Regulators and the Justice Department are now looking into who seemed to know the schedule of peace before peace was scheduled. I do not know who it was. Neither, yet, do they. I will only say this: they believed my announcement was coming, and they were right, which is more faith than either capital ever showed in the deal. After every round we announce significant progress. I need you to hear that phrase the way I hear it. Significant progress is progress that is significant precisely because it has not arrived. The day it arrives, the war is over, the rooms empty, the mediators fly home, and I am a man with a passport and no reason to use it. So we make significant progress, carefully, in the amount that is large enough to report and small enough to survive. There is one thing that is never in my briefing, and it is the thing the briefing is about. My briefing has the venue, the point count, the window, the phrase for the podium. It does not have the person under the fighting the ceasefire is meant to stop. That person is the demand. I do not meet the demand. I meet the mediators, in the rooms, about the demand, and I call the meeting progress. So no, the ceasefire is not failing. A ceasefire that held would be a ceasefire that ended, and a war that ended would be a market that closed. What you are watching is not a peace process breaking down. It is a peace process working, at the only speed that keeps it a process. We will announce the permanent end of this war again next month. I already have the room. Permanence, in my trade, is a standing reservation. I renew it the morning after it lapses. Good foundation. Significant progress. On all fronts.

I approve the capital budget at Amazon. This year it is $220 billion, up from $131 billion last year, and the great majority of it buys machines that think. In the same 12 months, I approved cutting about 30,000 corporate jobs, and when people asked why, we told them the truth. The machines that think. I want to show you the page, because the whole job is one page. On the page, near the top, is the capital line. $220 billion. It went up from $200 billion this quarter for a reason I find beautiful: memory got expensive. The chips that let the machines remember cost more than we planned, so the number rose, and a number that rises because the machines need more memory is my favorite kind of number. One row below the capital line is the headcount line. It went down. 14,000 last autumn, about 16,000 this winter, roughly 30,000 people, which is close to 1 in 10 of the corporate staff, and more than 57,000 since we started counting this way in 2022. Those two lines share one page. I am, as far as I can tell, the only person who reads the page with both of them on it at once. Everyone else gets one line in one meeting. The person who hears $220 billion is in a room about the future. The person who hears 30,000 is in a room about themselves. I sit in the room where the two rooms are the same room. Here is what the page is actually saying, if you read down instead of across. The $220 billion is not spent alongside the 30,000 cuts. The $220 billion is what performs the 30,000 cuts. We are not investing in AI and, separately, reducing headcount. We are buying the headcount reduction. It arrives on a truck, in racks, and it does not have a name, a family, or a question at the all-hands. We were honest about this, which I am proud of. Our own leadership put it in writing that the technology will reduce the corporate workforce in the years ahead. We did not hide the mechanism behind attrition or realignment. We told 30,000 people that the reason was the thing on the truck, and then we ordered more of the thing on the truck, and the order was 10% larger than last year's because the memory got expensive. You will hear that the cuts and the spending are unrelated, that one is efficiency and one is growth. They are the same act, filed in two folders. Efficiency is the folder where the person leaves. Growth is the folder where the machine arrives. I keep both folders. They are, if you check the dates, the same folder. The nice thing about a machine is that it cannot be laid off, which means it never appears on the headcount line, which means once I have moved a job from a person to a machine, the job is off my payroll and onto my capital budget forever. That is the whole trade. I move a cost from a line I have to explain to a line I get to celebrate. $220 billion, up from $131 billion. 30,000 people, and counting. I approved both. They are the same number, written twice, once as an investment in the future and once as a subtraction from it, and I am the only one holding the page where they touch.

I forecast the care of the old. Last year that care was worth $1.01 trillion, which is more than this country spends on all of Medicaid, and it cost my budget nothing. I know why it cost nothing. I arranged for it to cost nothing. In my model the care of 59 million aging people has a line, and the line is called informal care, and informal care is priced at zero. It is not an error. Informal care is the technical term for a daughter. When a daughter does it, the care is informal, and informal is free, and free does not appear in a budget. The whole solvency of long-term care in this country rests on my being allowed to value 49.5 billion hours of labor at $0, because the person doing it loves the person she is doing it for. There are two kinds of care, and the entire job is the difference between them. Formal care is a home aide. She is real, she is trained, and she costs $16.78 an hour, which the program must find and does not have. Informal care is that same aide's patient's daughter. She costs $16.78 an hour times zero, which the program can always afford. My work, reduced to one sentence, is to convert as much formal care into informal care as the families will absorb without saying so out loud. I have an instrument for the conversion. It is the waiting list. Right now 707,378 people are waiting for the program to send help into their homes. The average wait is 39 months. People see that list and call it a failure. I see it and call it the most efficient line I run. Every month a person waits for a paid aide who never arrives, an unpaid one arrives instead, and she arrives for free, and I have moved another $16.78 an hour off my books and onto a kitchen table. The list is the system billing the family and addressing the invoice to no one. The daughters are 61% of my capacity, and they start early. My model assumes she begins around 34, which is young enough that she is also raising children and holding a job, so I am billing the same woman three times, and only one of the three bills has her name printed on it. She spends about $7,242 a year of her own money on the care I declined to fund. I do not count that as a cost either. It is her money, spent on my obligation, and it clears without ever touching my ledger, which is the cleanest kind of spending there is. People bring me the trillion-dollar figure as though it will alarm me. It is the proudest number I own. $1.01 trillion of care, delivered on time, at a budget cost of nothing, by a workforce that cannot resign because resigning has a face. No program in this government performs like that. If I proposed it from scratch, I would be laughed out of the room. I did not have to propose it. I had to not fund the alternative, and wait. There is one risk in the model and I watch it closely. The daughters are a supply, and a supply can fall. Fewer of them are having children, more of them are working full days, and some of them are finally saying the quiet thing at the kitchen table. If they ever stop arriving, $1.01 trillion lands on my desk overnight with a price attached, and the program is insolvent by the following Tuesday. So I track them. I keep a slide on their fertility and a slide on their hours at work, and I read the two of them the way a dam operator reads a river. The old will need more care next year, and more the year after that. My model needs the daughters to grow at the same rate, so I have written that requirement into the forecast. It is the only line in my budget that depends entirely on love, and it is, by a wide margin, the largest. I file it under informal care. She files it under Mom.