In May 2023, the US Treasury announced plans to bring back regular Treasury buybacks starting the following year.
The idea was fairly simple: buy older, less liquid bonds and improve how the Treasury market functions.
On May 29, 2024, they actually started doing it.
Then in July 2025, the program became more focused on the long end. Buybacks in the 10–20Y and 20–30Y sectors started happening more frequently.
Yesterday, they went one step further.
Starting September 9, the maximum size of those long-end liquidity support buybacks will at least double from $2B to $4B per operation.
So over the last three years:
2023: announce the buyback program.
2024: start using it.
2025: use it more often in long-dated Treasuries.
2026: double the size of those operations.
I want to be clear about one thing here because I know where the conversation usually goes:
This is not QE.
The Fed isn’t printing money to buy Treasuries. Treasury is managing its own debt market, buying older securities and trying to keep liquidity from becoming a problem in parts of the curve.
But saying “this isn’t QE” doesn’t make the progression irrelevant either.
The US has more and more debt to finance. Long-term yields have been under pressure. And over the same period, Treasury has gradually become more active in supporting liquidity at the long end.
Now put gold next to that timeline.
I’m obviously not saying Treasury buybacks are why gold went from around $2,000 to where it is today. There are plenty of other things behind that move.
But I do think both are part of the same bigger story.
The harder it becomes to finance growing deficits without creating stress somewhere in the bond market, the more interesting an asset with no issuer, no maturity and no counterparty becomes.
Treasury can improve liquidity. It can change issuance. It can buy back older bonds and make the plumbing work better.
What it can’t do with any of those things is make the debt disappear.
The idea was fairly simple: buy older, less liquid bonds and improve how the Treasury market functions.
On May 29, 2024, they actually started doing it.
Then in July 2025, the program became more focused on the long end. Buybacks in the 10–20Y and 20–30Y sectors started happening more frequently.
Yesterday, they went one step further.
Starting September 9, the maximum size of those long-end liquidity support buybacks will at least double from $2B to $4B per operation.
So over the last three years:
2023: announce the buyback program.
2024: start using it.
2025: use it more often in long-dated Treasuries.
2026: double the size of those operations.
I want to be clear about one thing here because I know where the conversation usually goes:
This is not QE.
The Fed isn’t printing money to buy Treasuries. Treasury is managing its own debt market, buying older securities and trying to keep liquidity from becoming a problem in parts of the curve.
But saying “this isn’t QE” doesn’t make the progression irrelevant either.
The US has more and more debt to finance. Long-term yields have been under pressure. And over the same period, Treasury has gradually become more active in supporting liquidity at the long end.
Now put gold next to that timeline.
I’m obviously not saying Treasury buybacks are why gold went from around $2,000 to where it is today. There are plenty of other things behind that move.
But I do think both are part of the same bigger story.
The harder it becomes to finance growing deficits without creating stress somewhere in the bond market, the more interesting an asset with no issuer, no maturity and no counterparty becomes.
Treasury can improve liquidity. It can change issuance. It can buy back older bonds and make the plumbing work better.
What it can’t do with any of those things is make the debt disappear.