A Buyback Letdown
The Weekend Edition is pulled from the daily Stansberry Digest.
The Treasury ups its buybacks (again)...
On Wednesday morning, the Treasury Department announced that it will buy back up to $6 billion in 10-year and 20-year Treasury bonds at a time. That's triple the typical amount.
The operation started the following morning, when the Treasury made its first purchase since announcing its increased buyback program last month.
As we wrote in the August 19 Digest...
Just yesterday, we wrote about the multidecade high in the 30-year Treasury yield. Between weak demand at last week's Treasury auction, ever-growing government deficits, and the ongoing conflict with Iran, investors have been staying away from long-dated U.S. debt.
This morning, the government stepped in to provide support...
In a press release, the Treasury Department announced that it will at least double the size of its long-dated Treasury (10-year to 30-year) bonds-buyback program.
Starting on September 9 (and running until the day after midterm elections), the Treasury will buy back $4 billion of U.S. debt at a time. That's double the current $2 billion rate.
At the time, Treasury Secretary Scott Bessent said that the buyback may be larger than the $4 billion announced. And he said the Treasury may use its $1 trillion general account to buy back the bonds.
But $6 billion wasn't enough for the market...
Yields spiked after the Treasury's announcement, with the 10-year yield rising to a high of 4.85% – its highest level since October 2023. The 30-year yield is closing in on its August highs, when the Treasury first stepped in.
The buybacks are an attempt to lower yields and prop up demand for longer-dated Treasurys. Wednesday's move is the opposite of what the Treasury Department wants.
So what's going on?
Well, Wall Street may have gotten ahead of itself. Before the announcement, Bloomberg – citing market strategists – speculated that the buyback figure could be as high as $10 billion.
When you take that into account, "only" $6 billion was a letdown. The market may be trying to push yields higher to see how far the Treasury will go to support U.S. bonds.
Bessent says, "I am the house now"...
In his previous career, Bessent worked alongside George Soros to short the British pound and "break the Bank of England." The trade famously made their firm more than $1 billion.
Now, Bessent is on the other side of the currency trade – working with governments to stop things from "breaking."
That goes beyond the U.S. bond market. As Brett Eversole explained in the August 12 issue of True Wealth Systems' "Review of Market Extremes"...
On July 31, the [U.S. Treasury] stepped in to buy up Japan's currency... stopping its rapid decline. And documents show the U.S. could buy up to $10 billion yen to support the currency.
It worked... The yen has surged from its July low of 163 yen per dollar to about 154 yen per dollar today – the strongest level for Japan's currency since February.
But plenty of traders are still shorting the yen. As Barchart showed in a post on X, short positions in the yen "remain near the largest in history." And Bessent has warned that traders who continue betting against the yen do so at their own risk. As he said in Texas this week...
I am the house now. So when we intervene with the Japanese yen, I have pretty good insight into what the Japanese, what the Bank of Japan, is going to do, what Japanese policymakers are going to do.
The house always wins. In short, Bessent says he knows what's coming.
Markets are expecting another rate hike from the Bank of Japan at this month's policy meeting.
U.S. intervention, coupled with another rate hike from the Bank of Japan, could be the end of the "carry trade."
The AI Boom Just Hit a New Milestone
In a post on X last weekend, Nvidia (NVDA) CEO Jensen Huang congratulated AI startup OpenAI on achieving "AGI" – artificial general intelligence. Put simply, AGI is the step in the AI boom when models match or surpass human performance.
According to the ARC Prize Foundation, which measures progress toward AGI, OpenAI's recently launched GPT-6 Astra model has achieved AGI.
As ARC Prize Foundation President Greg Kamradt said...
On ARC-AGI-3, Astra surpassed our human action-efficiency baseline on 96% of levels, effectively reaching human parity on the benchmark. Not only is this the best model we've ever tested, but it also represents a meaningful step change in frontier-model performance – not only in its ability to navigate and solve novel environments, but also in how efficiently it learns to do so.
OpenAI also claims to have solved the Navier-Stokes problem – a previously unsolvable mathematical problem, according to CNBC. It took 10,000 OpenAI agents about 88 hours to solve the 90-year-old problem.
With this latest release, OpenAI is catching up to competitors like Alphabet (GOOGL) and Anthropic... or even taking the lead again.
And the news couldn't come at a better time, with both OpenAI and Anthropic planning to go public soon (Anthropic possibly as soon as next month).
But it comes at a cost...
In a now-viral post on X, former Anthropic and OpenAI researcher Jacob Coxon announced that he had left Anthropic and said the two AI startups are not "acting responsibly" and are "gambling with our lives."
It's not just a case of a disgruntled former employee. In response to Coxon's post, Evan Hubinger – a current "Alignment Science Lead" at Anthropic – agreed, saying he sees a greater than 10% chance that AI will kill off civilization within the next 10 years.
The companies themselves have suggested this kind of threat before. In its August 2026 Risk Report, Anthropic acknowledged that its AI models are self-improving at a faster rate than it previously thought.
However, we doubt that "AI Causes the End of Civilization" will make it onto the Risk Factors pages of these companies' S-1 filings for their upcoming initial public offerings ("IPOs").
In SpaceX's (SPCX) S-1 filings, the closest we got was, "The global nature of our business poses risks with respect to unstable, malicious or arbitrary legal regimes and authorities."
Of course, AI companies can use this kind of press to talk up the power of their technology, too. It's also important to separate real risks from doomsday scenarios... As with many things, if you focus on the most extreme predictions, you might miss dangers that are more likely.
But with OpenAI's and Anthropic's IPOs looming, every AI headline will be under a microscope. And these concerns could take some of the shine away from the AI bubble.
All the best,
Nick Koziol
Editor's note: Whitney Tilson has built his reputation by getting ahead of the crowd. He was early to some of the biggest market moves of the past three decades – from the dot-com bubble to tech pioneers like Apple, Amazon, and Netflix. Now, he sees signs that the market's AI leadership could be shifting... and a new group of companies could take over the AI spotlight.