Circle Chief Economist Discusses: Entering Hyperliquid, Clear Legislation, and the New Federal Reserve Chair

Compile | Deep Tide TechFlow

Original text link:

Editor’s Note

In this episode of the podcast, Circle’s Chief Economist Gordon Liao впервые provides a systematic explanation of the market-structure logic behind USDH being replaced by USDC. Over the past year, USDC balances on the Hyperliquid platform have doubled. 90% of reserve earnings flow back to Hyperliquid to repurchase HYPE. Coinbase serves as the treasury deployment partner, Circle serves as the technical deployment partner, and stakes 500,000 HYPE.

Gordon also breaks down long-end U.S. Treasury yields. The current 30-year yield breaking above 5% is mainly driven by term premium. Meanwhile, stablecoins are quietly becoming the marginal buyers of U.S. Treasuries. In Q1 2026 alone, USDC’s on-chain settlement volume reached $2.1 trillion. Concentrated buying of short-term U.S. Treasuries by stablecoins effectively lowers the overall weighted duration of U.S. government debt, which may provide inverse support to long-end rates.

In addition, the show is also worth attention for its take on the sticking points in the CLARITY bill, and its disagreements on which layer captures the value of AI after the OpenAI lawsuit.

Key Quotes

USDH is replaced by USDC

At its core, this is a liquidity supernova event. As the most dominant on-chain perpetual contract platform, the collateral assets it uses will radiate across the entire chain economy.

The governance vote about eight or nine months ago selected a different reference asset. But as the platform grows and matures, it also needs to deal with more traditional institutions. Using high-quality, institutional-grade collateral assets is a key part of that.

Wherever there’s an opportunity to intercept TVL—whether it’s exchanges or prediction markets—everyone will look for ways to monetize this floating interest rate. Why should they give this money to a third party?

For Coinbase and Circle, this is a strategic move to neutralize an emerging competitor. Coinbase, as the collateral-management party, places itself at a key node in this new infrastructure.

Stablecoins’ multiple attributes

When it comes to whether stablecoins are truly a medium of exchange or a store of value, we can see they can be both at the same time. In payment scenarios, they are a medium of exchange; in this case, they are the carrier of capital liquidity and collateral liquidity. As the system’s scale expands and it becomes more institutionalized, the latter will become increasingly important.

Agents will probably want a money-market fund-like currency that continues paying interest while idle. But the moment it’s time to make payments, they’ll want that money wrapped as stablecoins. Just the compliance paperwork for paying with securities is already unbearable.

OpenAI case and AI value capture

At the LLM layer, there’s almost no value capture. These AI Labs spend hundreds of billions to provide us with free services—in essence, they’re doing public service. The value of LLMs lies in model weights, which is IP.

Who controls the end users delivers the greatest value. Value mainly sits at the application layer, as well as in cloud business and AI deployment service providers. For example, companies like Accenture can earn quite well.

I think it’s a barbell structure. Aside from the distribution side, the other end is energy. Whoever can get nearly free energy and cheap computing power wins. Elon has an advantage there.

About the CLARITY bill

The compromise proposal between Thom Tillis and Angela Alsobrooks essentially separates the functions of money’s value storage, settlement, and accounting unit.

We are approaching Hillary’s Step—the final hard part before the summit. There are also issues with committee seats and ethics. The ethics hurdle will be extremely difficult.

From the very beginning, I felt this battle was bank-industry style, very Don Quixote-like. What exactly do you want to get out of it? Is it just stabbing each other with knives, or are you just delivering a big gift to asset management companies?

Long-end U.S. Treasuries and rates

Most of the upward momentum in the 30-year yield comes from the term premium. It’s currently about 80 bps, which is already quite high compared with the negative levels two years ago. This means the market is reflecting supply and demand, not expectations for short-term rates in the future.

The narrative that stablecoins are the marginal buyers of U.S. Treasuries contains more substance than the credit story people give. Their duration is very short, concentrated in short-term Treasuries and reverse repos. In practice, this releases the Treasury’s space to issue more debt in the short term. If calculated using dollar duration weighting, it amounts to reducing the supply of long-term U.S. Treasuries in the market.

Investors are saying: “I need more compensation to hedge against bigger inflation risk. That’s all.” They know the Fed is not inclined to cut rates.

Coinbase and Circle team up to take USDH

Austin Campbell (Host): Hello everyone, welcome to Bits + Bips. In every episode, we explore how crypto and macro collide in basis points. I’m your host, Austin Campbell. Today’s guests are Lumida Wealth co-founder and CEO Ram Ahluwalia, CoinFund managing partner Chris Perkins, and Circle’s Chief Economist Gordon Liao. There are plenty of topics in the rates market and the news today, and I’m especially looking forward to Gordon’s perspective.

Let’s start with Circle today. Coinbase and Circle basically ate USDH. USDC will be crowned as the quote asset aligned with Hyperliquid (an on-chain perpetual contract DEX). Native Markets, which won the governance contest eight months ago, is now being acquired by Coinbase. USDH holders will redeem for USDC during the migration period. Coinbase will become the official reserve fund management/deployment partner for USDC on Hyperliquid. Circle will handle USDC’s technical access and the operating infrastructure on Hyperliquid, and stake 500,000 HYPE to send them toward validator seats. 90% of reserve earnings will flow back to Hyperliquid, likely through a support fund used for HYPE repurchases.

Let’s do a rough calculation. Right now, Hyperliquid has about $5 billion in USDC. With returns of under 4%, the annualized figure is approaching $200 million. Most of this money will flow into Hyperliquid; Coinbase takes a cut; Circle gets a new USDC deployment base and continues to chase Tether’s scale.

The bullish logic is that with deeper order books, less exchange/replacement loss from swaps, faster deposits and withdrawals, and better market maker support, HYPE becomes more attractive in terms of the platform’s fees, staking, and Builder activities. Bitwise is also applying for a spot HYPE ETF. The bearish camp—for example ZachXBT—worries that if Hyperliquid’s core collateral and quote assets and liquidity become increasingly dependent on USDC, then the entire system hands over some critical lifelines to Circle/Coinbase/regulatory commands. There are also governance issues with Native Markets. Chris, as an investor in this space, what’s your view?

Chris Perkins: I think this is one of a series of moves we’ll see next. The keyword is “net interest income.” If you step back and look at the traditional exchange model: you earn ticket fees; for every transaction, you take a small cut. Typically, clearing doesn’t make money, although in our circle it becomes a new business line, and sometimes you earn a bit from data. But the real big money is net interest income.

In traditional finance, what happens is: clients give you dollars as collateral. You give the dollars to the clearinghouse. The clearinghouse invests the dollars, keeps a large chunk, and gives you a small share back. That is your net interest spread. This is the core building block of any exchange business model. Many decentralized applications have ignored this and handed that handsome income away for free. Now they realize they need to take it back.

I can tell you that anywhere it can intercept TVL—exchanges, apps, prediction markets—will try to monetize this floating interest rate. Why should it be given to third parties?

If you’re looking from the exchange angle, this is the bullish logic. Hyperliquid went up after the news, because this circle is completed (a pun on Circle). You solve the net interest income problem and the rewards flow back to token holders. If you’re on the Circle/Coinbase side, you’re also a winner—of course, the details are in the terms, like how long the hook period is, how often rates are renegotiated. I don’t know whether Gordon can disclose that. But what you get is a widely distributed stablecoin—USDC is substitutable. The more it circulates, the more likely it is that end users accept it as a payment method.

So USDC also wins. Perhaps in the future both sides will adjust the economic terms. Hyperliquid wins big by locking in net interest income; Circle gets greater adoption, bigger scale, broader distribution, and the hoped-for incremental utility. I think it’s a win-win.

Austin Campbell: Gordon, I want to throw a question to you. I know Circle well too, and in today’s market USDC has many different facets. From a market-structure perspective, Americans are used to viewing money as layered: the money you use to buy coffee is not the same as the money you use to settle derivatives. But now we’re starting to see USDC being used for many things, and substitutability is getting stronger. How do you see this from Circle’s perspective? And zooming out to your economics background, how does this change market structure?

Gordon Liao: A few observations. First, we’re witnessing the overall maturation of infrastructure. Hyperliquid is today’s dominant on-chain perp (perpetual contract) platform, and it has grown a lot in size. USDC’s balance on that platform has roughly doubled year over year.

The governance vote about eight or nine months ago did choose a different reference asset. But as the platform grows and matures, they also need to deal with more traditional institutions, and using high-quality, institutional-grade collateral assets is key to that. Choosing USDC is recognition of its underlying security and the underlying 1:1 reserve commitment.

As Chris said, this is a win-win and a liquidity supernova event. As the most dominant perp platform on-chain, the collateral assets it uses will radiate into the entire chain economy. So it’s a huge liquidity event. It will also encourage the use of USDC and the other related infrastructure.

We deployed USDC to Hyperliquid back in September last year, together with CCTP (Cross-Chain Transfer Protocol). So it has been there for a while. But this time is a good “reconfirmation” event.

When it comes to whether stablecoins are primarily a medium of exchange or a store of value, we see they can be multiple things at once. In some scenarios, they are a medium of exchange for payments. In other scenarios, they are a carrier of capital liquidity and collateral liquidity. As the system scales and becomes more institutionalized, the latter will become increasingly important.

Similar trends are also clearly visible in settlement volume. In our recent Q1 report, USDC’s on-chain settlement volume was $2.1 trillion. This reflects infrastructure expansion: liquidity on the largest platforms—whether centralized or decentralized—is improving.

Austin Campbell: Following that thread, USDC’s circulation is actually highly tied to Coinbase. Many Coinbase products are built on USDC, such as debit cards, credit card payments, and enterprise payments. Now we’re also using it as a core asset for exchanges like Hyperliquid. Ram, from a market perspective, does this make you more bullish or more bearish on Coinbase and Circle stock?

Ram Ahluwalia: It’s good news for everyone, especially for Hyperliquid. For Coinbase and Circle, they succeeded in neutralizing a new emerging competitor. Coinbase, as the collateral manager, placed itself at a key node in the new infrastructure. That is a very strategic move.

For Hyperliquid, keeping 90% of the revenue is the payoff for its results over the past few years. We discussed Hyperliquid a few weeks ago—it’s one of the assets you’d want to hold during this cycle. Coinbase moving first and taking action early is very forward-looking, because Hyperliquid is becoming the core of decentralized trading venues. Circle has gotten a sizable recurring net interest income. So it’s one of those situations where everyone wins, and the benefits are especially great for Hyperliquid.

This also ties back to another topic we discussed earlier: distribution will ultimately drive most of the revenue in this system. Gordon, you also said that Hyperliquid is emerging as a perp DEX in crypto. As all these parties come together, the essence is recognition of distribution and user positioning. This main line will keep showing up when assessing who wins and who loses.

OpenAI lawsuit

Austin Campbell: Speaking of users and winners—today Elon Musk lost, Sam Altman won, at least in the first round. An Oakland federal jury unanimously rejected all of Musk’s claims in under two hours. The core of the ruling is the three-year statute of limitations. The jury found that Musk already knew about OpenAI switching to a for-profit model in 2021, but he only sued in February 2024.

He originally sought $134 billion in “ill-gotten gains,” and to remove Altman and Brockman from leadership, citing a 2025 for-profit restructuring. But substantive issues—including breach of a charitable trust and unjust enrichment—were not decided. Musk’s team has said it will appeal. Wired commented that during the trial, both sides portrayed each other as self-interested. Neither side looks clean. The subsequent interpretation is that, at least during the transition period, OpenAI might be able to go public.

There are a few reactions on X worth reading. Structural skeptics say that legally, this is a big win for OpenAI, but the bigger issues of politics and institutions remain untouched. What does it mean when an organization founded on a mission of “non-profit, human-first” gains public legitimacy, and then becomes one of the world’s most valuable commercial platforms? News24 says that a non-profit machine created to benefit humanity was forcibly turned into a closed, profit-driven machine backed by Microsoft. The trial did reveal breaches of commitments around openness and safety. Chris, what’s your take?

Chris Perkins: It seems the statute of limitations has run out, and that’s pretty straightforward. I don’t know how Musk’s lawyers will appeal, but they’re smart—they’ll probably find a way.

At this point, Ram would normally say something negative about OpenAI—he tends to blow things out of proportion. Before we let him, the bigger issue in crypto is that, due to regulatory pressure over the past four years, many foundations have been designed as non-profits, and there are also Labs. I’d like to have a clear legal precedent to sort out the relationship between foundations and Labs. Right now, in many protocols, people are confused about who is responsible for what, and who is who.

I’m not saying foundations have no value. Foundations absolutely have non-profit ideals that can push things forward, like Ethereum’s cryptography research. But many foundations may have been created to seek protection when facing a very aggressive regulator. So this case could have far-reaching implications for crypto. Sam is also increasingly getting involved in this circle.

Ram Ahluwalia: Chris, you basically handed me the ball. I didn’t expect this case to generate any big waves, and indeed nothing much happened—so it’s a small issue.

The tech world has heroes and villains. I put many people on the hero side and some on the villain side, but that doesn’t mean they haven’t created value. Sam’s problem is that he has a history of signing illegal agreements at the contract law level. He even played this same move with Microsoft—after signing an agreement with Amazon, he then renegotiated Microsoft’s contract. As the cost, Microsoft got very favorable terms, and OpenAI got the funding it needed. Of course, Microsoft also wanted to realize a 10x return by providing capital.

In my view, Sam is clearly a villain. He was kicked out by a board he appointed himself. And unexpectedly, while he was suspended, his team left. Also, an employee died under suspicious circumstances while he was still in power.

Chris Perkins: You’re a bit extreme here.

Ram Ahluwalia: No, no—this is a factual statement. “An employee died under suspicious circumstances.” That’s an accurate statement, and the circumstances are indeed suspicious.

In short, this field has heroes and villains, and I put him in the villain bucket.

Austin Campbell: Gordon, what do you think?

Gordon Liao: More broadly, AI competes intensely at every layer. What we’re seeing in court is one aspect of that competition. But if we think about it for the audience of people who care about both blockchain and AI, where are the opportunities? I think it’s about building the track for tomorrow—for Agents, for AI. That’s also what Circle has been doing: releasing an agent technology stack, and what we call ARC, an “Economic Operating System.” I believe these will have enduring network effects, as strong as USDC’s network effects. So even if the base model layer is highly competitive, from a business perspective there are many opportunities elsewhere.

Chris Perkins: Competition is a good thing—we need more of it. Nobody is perfect, and I’m not either. It’s a brutal competition. I hope they keep innovating and keep creating value, and that the free market wins.

Austin Campbell: Building on that, let me ask a question. Today we’re fighting fiercely in court for OpenAI, because private markets see OpenAI as one of the most valuable foundational model companies. But let me stress-test this: in the long run, will this trial be remembered as a small issue? The reason is: if distribution is where value ultimately accumulates, will the value stay with companies like OpenAI and Anthropic, or will it stay with the platforms that deliver models to users and collect tolls?

Ram Ahluwalia: Definitely the latter. At the LLM layer, there’s almost no value capture. These AI Labs spend hundreds of billions to provide free services for us—essentially public service. But Microsoft owns OpenAI’s IP. In six years, they can dispose of the stock, but the IP is forever. They can do anything with the IP—even put it on the internet. I’m not saying they will, but the value of LLMs is in the model weights, which is IP. It will end up in the hands of a direct competitor.

So I support AI Labs raising more money to invest in humanity’s future. But their business model doesn’t capture value. Meta brought in an “A Team” (Ram referenced the 1980s TV series A-Team and the Airwolf pun). Their new model is quite powerful, and they’re throwing a lot of money at it on NVIDIA GPUs. So this race is still early.

Anthropic is in a leading position and its revenue growth is fast. Today I also saw Michael Dell from Dell reveal they just signed 1,000 enterprise customers. We’ve moved from a world where only hyperscale data centers and AI Labs throw GPUs at things, into a real phase of commercial deployment—and it’s still early.

Who controls the end users delivers the greatest value—mainly at the application layer, cloud business, and AI deployment service providers like Accenture.

Chris Perkins: I agree with the logic that distribution matters. But I see it as a barbell. The other end is energy. Whoever can access nearly free energy and cheap compute power wins. Elon has an advantage there. The science of getting energy almost for free in space isn’t easy. But if anyone can do it, it’s the person who can send things to space worldwide. That’s Elon’s unfair advantage at the bottom of the stack. But the front end is still distribution.

Ram Ahluwalia: We still haven’t seen Apple truly bring out its solution. How many times has Apple entered the race late and then ruled everything? Recently there’s been some internal turbulence—this company is worth watching.

Austin Campbell: I want to say Apple is an interesting story here. When its “Mag 7” peers are spending CapEx on model training, Apple seems to be saying: We are a vertically integrated hardware company end to end—from iPhone, MacBook to servers and Mac Studio. We will become the endpoint for your model distribution, and we will extract tolls. Just look at what they do with the App Store, and what you see from ecosystem bundling makes it clear.

By the way, Chris, this also goes back to your identity obsession—Apple is one of the few big tech companies that does “acceptable” privacy protections. They’re more trusted on these issues. So the division I see now is: do you want to build your own AI, or do you want to deploy other people’s AI and charge tolls? The latter makes you the distributor.

Gordon, now you’re at a “money” company, and you’ve also seen plenty of cases where Agentic Commerce and finance companies entangle AI. How do you think this relates to the modernization of the U.S. financial system and adoption of new products? Reminder for U.S. listeners: Asia already had 24/7 real-time full payment systems from the late 1990s to the early 2000s. We’ve been behind by two decades. Will this accelerate the update of the entire financial economy—not just the AI segment?

Gordon Liao: Absolutely. Right now, most trading is still mediated by humans. But many forecasts believe that machine-to-machine, machine-initiated payments will dominate. Today, big LLM model companies are large and important. But model changes happen very fast, and even open-source models don’t fall too far behind. So value will increasingly flow into the tracks where commodities, hardware, and agentic commerce occur.

For example, small payments—recently we released a small payments protocol. It’s a marketplace where agents can browse, find each other, and use the best tools, so that even if nobody is present, machines can run machine-to-machine commerce. All of this will be built on blockchain tracks. We recently used ARC, which is a huge growth area, and it will integrate closely with finance. When we talk about CLARITY later, it will also come up—its relationship to balance-sheet-based intermediaries versus activity-based finance.

Austin Campbell: Here’s a counterpoint. I often hear that agent payments will move onto blockchain tracks, but I also think they’ll use traditional tracks. Agents getting a credit card is easy. I think the winners will be those who can connect flows seamlessly across different systems. That’s why I’m paying attention to combinations like Coinbase + Circle, or the products Fidelity is rolling out. They already have money market funds and cash management products, and now they also need to issue stablecoins.

Agents seem to have far less “loyalty” than human consumers, but they’re good at optimizing among different flows. Not all flows use the same framework—sometimes it’s on-chain payments, sometimes it’s card swipes, sometimes it’s bank accounts. I guess the winners in agentic commerce are those who can integrate all of these seamlessly. In theory, things trapped in purely on-chain or purely off-chain setups might lose to cross-chain on-chain-off-chain integrators.

CLARITY bill: reaching “Hillary’s Step”

Austin Campbell: The Senate Banking Committee passed the Digital Asset Market Clarity Act with a bipartisan vote of 15 to 9 and sent it to the full Senate. To enter formal consideration on the floor, it needs 60 votes. There’s also an unresolved question: after the bill reaches the full Senate, will there be further amendments?

The bill’s core includes: decentralization testing, the division of responsibilities between the SEC and the CFTC, and which agency oversees which kinds of tokens. Everyone agrees the bill isn’t perfect, but at least it’s a start.

First, let’s talk about two points. One is the controversy over stablecoin yields. Consumers, retail users, and the crypto industry call it “works as designed,” while the banking industry lobbying group still calls it a “loophole.” Senators Tillis and Alsobrooks reached a compromise: strictly defined passive yield is banned, but “activity-based rewards” are allowed. The American Bankers Association (ABA) is not happy, but other parties are somewhat satisfied, and senators also say “no more renegotiation.” Let’s talk about it: assuming Tillis and Alsobrooks are telling the truth and there won’t be renegotiation, if the bill passes with the current terms, what do you think?

Gordon Liao: I’ll start us off. Money itself has multiple attributes: it’s both a settlement tool and a store of value. In a sense, it separates the store-of-value function, the settlement function, and the accounting unit function.

This also echoes a bigger trend in the financial intermediation space. Traditional intermediaries rely very heavily on balance sheets, which is why there are banking rules that focus on stress-testing balance sheets. In that world, making a big balance sheet is key, paired with corresponding regulation. But if you look at the evolution of on-chain finance, a fairly large portion of it is activity-based. It’s not about balance-sheet size; it’s about a set of activities under smart-contract-based intermediaries.

This compromise hits the boundary between old and new very well—moving from the traditional, balance-sheet-heavy intermediary world to the new, smart-contract- and Agent-led intermediary world. Players who focus on activity-based rewards, activity-based services, and new intermediary forms will have a lot of opportunity.

Chris Perkins: First, I want to thank Senator Tim Scott and Senator Loomis. The Republican leadership did an excellent job. I also want to particularly mention Senator Gallego and Senator Alsobrooks. Gallego is a Marine; when I was in Ramadi, he fought in Haditha. He has courage. He helped make the bill bipartisan at the committee level—very good.

We’re now approaching Hillary’s Step (the last obstacle before the summit). Issues with committee seats remain. Ethics issues remain. And ethics will be very tough. Then there’s the fact that the banking industry is unhappy. Their lobbying is still pulling the national-security line today, probably because they’re unhappy with other provisions. So getting close to the finish line will become extremely difficult.

But I still think the bill will pass in the end. I want to hear what you all think, because we’ve had disagreements for a while.

Ram Ahluwalia: I think it’ll barely pass. What did Trump do? He posted a tweet about the CLARITY bill, saying he’d put his chips on it. With midterm elections coming up, there’s no upside in letting it fail. I still think it’ll barely pass.

Austin Campbell: I’m more skeptical than both of you. Getting to the full Senate is a very positive signal; regardless of what the previous probability estimates were, it’s now higher. But Chris, at the moment nobody has given me a truly credible “ethics” solution. I see two possible paths. One is that in the short term it looks good but in the long term it’s very bad—CLARITY passes along party lines in the House and Senate. That’s possible under the current frameworks. But if it’s only passed by Republicans, it might play out like the Affordable Care Act: once the other side comes to power, they start dismantling it. Pushing a transformative piece of legislation through when it’s one-sided along party lines in U.S. legislative history has never been a good outcome.

The other path is that the bill directly hits the ethics reef and sinks. If it dies, it dies there. Every other issue—including the banking industry’s disputes about yields—has solutions. Many of the arguments are essentially industry special pleas, which are against the interests of ordinary consumers, the U.S. economy, and national-security agencies. And the latter are increasingly interested in digital assets and hold more positive views. On ethics, after talking to people on both sides, it’s still an area where I have doubts. So I’m hitting the brakes.

Ram Ahluwalia: Austin, what exactly are the sticking points on ethics?

Austin Campbell: The core is whether Democrats would vote for a bill that doesn’t force Trump’s family to divest from World Liberty Financial, meme coins, and other interests. Would Republicans send a bill that forces the president to do so? That’s the middle-ground breakpoint. I don’t see a clean solution. But Chris is right—once it reaches the full chamber, it will get weird. It has a non-zero chance to move forward, and maybe some compromise emerges that’s unrelated to crypto, like swapping other chips.

Ram Ahluwalia: It’s very clever to distinguish between interest payments and “activity,” and in principle I like it. The question is: how much room do activity-based regulations have to accommodate edge cases?

Chris Perkins: We’re now living in the “post-Chevron deference” era (referring to the Supreme Court overturning Chevron, meaning courts no longer automatically defer to regulators’ interpretations). Previously, when rules weren’t clear, regulators could interpret them as they pleased, filling in gaps where the text was ambiguous. Now legislation has to be more rigid and more prescriptive, and the result is that bills themselves become worse. Then any dispute goes to court. In a sense, we do need to let Chevron fade out. But it also had benefits in the past. I’m not saying bring it back—I’m just saying complexity has increased.

Austin Campbell: Let me add another angle. In essence, this is a Gordian knot, rooted in the structural problems of U.S. financial regulation. If you want to prevent money-market funds from paying interest, you’d have to rewrite the 1940 Investment Company Act—under the law, they’re required to pay yields to clients. They can’t keep the money inside the instrument to compound. As long as we have tokenized securities on one side, stablecoins on the other, and reserves that resemble tokenized securities, this door can’t be shut. The issue is form rather than substance. Otherwise you’d have to rewrite the Banking Act and the 1940 Act. I don’t think Congress has the will to do that.

They even almost couldn’t confirm Warsh—he was just put into a chair to do a specific job. Are we expecting them to rewrite the entire structural foundation of U.S. securities regulation? Not going to happen. So from the beginning, I felt the banking industry is fighting this Don Quixote-style battle—what exactly do you really want? Isn’t this just stabbing each other and delivering a big gift to asset managers?

Chris Perkins: Senator Gillibrand is now one of the most watchable people. She supports crypto and has been deeply involved since the earliest stages of legislation formation, but she’s extremely firm on ethics. If someone can reach a deal with her, it will be very influential.

Bond vigilantes vs. the Fed’s new chair, Warsh

Austin Campbell: Warsh is the most closely divided Fed chair in modern history. The first FOMC meeting is in mid-June, but a few hours after his confirmation, a $25 billion 30-year Treasury auction broke above 5%. The 30-year yield touched 5.12% intraday. For the first time since the 2008 financial crisis, we saw the “5-handle.” The 10-year is currently 4.59%, and the 2-year is 4.08%. The CME FedWatch tool shows a 50% probability of rate hikes later this year, completely reversing the previous narrative of rate cuts. Of course, there’s still room to debate how to interpret this data—it’s not a simple one-dimensional rate-hike line.

Ed Yardeni, who coined the term “bond vigilante,” says they’re setting policy now and the Fed might be forced to hike in July. Vincent Ahn of WisdomTree says Warsh wants to keep the option of cutting rates on day one, but the bond market has just taken that option off the table. Morgan Stanley says rate cuts will be pushed to 2027. Ryan Swift of BCA says if Warsh takes a dovish stance amid this bloodbath, inflation expectations could become unanchored, and the Fed could lose control of the long end.

Some people think this is a good thing. Phil Blacanto of Reuters Breakingviews says the evaporation of expectations for rate cuts could constrain an over-interventionist Fed, which might be positive. But interest rates are a complex space and are often misread in traditional markets. Gordon, as an economist, a market observer, and someone with Fed background, what do you think the market is saying now?

Gordon Liao: I’ll answer from a background perspective. I used to work at the Federal Reserve Board in Washington. In a role like that, when you see rates move, the first thing you ask is: is it the term premium moving, or is it expectations for short-term rates moving?

If you use the classic ACM decomposition, most of the rise in the 30-year yield is driven by the term premium. The term premium is currently around 80 bps, already quite high compared to two years ago when it was negative. The term premium basically reflects supply and demand, while expectations for short-term rates reflect the market’s view on whether the Fed will hike.

The rise in yields mainly comes from the term premium, meaning the market is reflecting supply-and-demand related factors. On the demand side: persistent fiscal problems and fiscal expansion; in the long run, a decline in confidence in the Fed’s ability to control inflation; and a weakening of foreign demand—the changes in international balance of payments over the past year have been significant.

On the supply side, it’s also interesting that Warsh, the chair-to-be, is a supporter of shrinking the Fed’s balance sheet. That implies reverse QE. This shows up alongside the pressure on long-end yields you’ve been seeing.

Another narrative is that stablecoins are the marginal buyers of U.S. Treasuries. I think this story is more solid than the credit story people give. Not only is the stablecoin supply increasing, but also the term structure issue. The assets held by stablecoins have a very short duration. That creates space for the Treasury to issue more debt in the short term.

If you compute with duration weighting, this actually implies a significant shift in dollar duration. It reduces the supply of long-duration dollars in the market, which could in turn ease some of the current upward pressure on rates. So these things are connected. Don’t just look at rates as a single number. Break it into term premium and expected short rates. Then also consider balance-sheet level—expected changes in the Fed’s balance sheet—and private-sector demand.

Austin Campbell: One of the points you just made is often ignored: it’s not just who’s buying Treasuries, but more specifically, which maturities they’re buying. Under the GENIUS Act constraints, stablecoins prefer short-term Treasuries. Even if they don’t strictly prefer them, if they do reverse repos, they can use Treasuries as collateral. But longer-maturity Treasuries have bigger discount rates, so there is still a preference by maturity.

At the same time, the biggest buyers of 30-year Treasuries are insurance companies and sovereign wealth funds. Their preferences are changing. Sovereign funds are reducing long-term Treasury holdings, possibly with a lot of geopolitical considerations mixed in. Insurance companies’ demand is tied to the population demographic curve of the countries they serve. The Baby Boomer generation is gradually exiting, and the next major group is Millennials. That reshapes the demand curve for 30-year bonds, likely lowering demand.

So I’m watching the term premium. Gordon, you said 80 bps is relatively high recently, but compared to true historical levels it’s actually still low. Historically it can reach 150 bps or even higher. The term-premium curve itself is changing shape. That’s the part that gets ignored in the story. Ram, from an investing perspective, how do you view long-end bonds?

Ram Ahluwalia: First, I agree it’s driven by supply and demand. Investors are saying: I need more compensation to hedge against bigger inflation risks. That’s it. They know the Fed isn’t inclined to cut rates, so someone has to blink—how that happens is through rates moving higher.

You can see the inflation rebound in oil prices affecting core goods, gasoline, and so on. It’s kind of ironic that Warsh’s confirmation vote was so one-sided in favor. Elizabeth Warren’s core test is: will he cut rates? He wants to cut rates, so this vote is more of a political signal.

Overall, I expect rates are topping out. If that’s true, then interest-rate-sensitive sectors that were hit—like insurance—will rebound. Insurance is a balance-sheet business, similar to banks. When rates rise, the value of bonds falls, and the discounted value of future liabilities also falls. So they’ve been under pressure.

But the most interesting change in the past two or three days is that long-duration, high free-cash-flow assets have started to rebound. Over the past couple days, major indices all fell by more than 1%, but long-duration, high-cash-flow assets rose. The stock market is telling you they believe long-end rates will fall. That’s kind of amusing—usually everyone treats the bond market as part of the stock market, but this time, maybe the stocks are right. Warsh’s confirmation is a kind of capitulation event. We still need that Ray Dalio blockbuster video about “Doomsday,” which hasn’t come out yet.

Chris Perkins: I’ll add one more point. Over the long term, some very deflationary pressures are hitting the economy. First AI, then energy prices, with the scale of investments in cheap energy being astonishing. Elon will push it into space, and the computing power that comes back from space will be unprecedented. These are long-term deflation factors. Warsh himself has also discussed this, especially AI.

The short-term issues are different. Sovereign wealth funds are selling Treasuries—some to decouple, some for liquidity. Oil prices are high for a reason. This government is different from any post-war government: first, it re-examines how inflation itself is measured; second, the collaboration between the Treasury led by Bessent and the Fed is unprecedented. I think this collaboration will lead to a more comprehensive policy response. That doesn’t mean Warsh isn’t independent—you can be both independent and collaborative, and I think that’s healthy. So I’m optimistic.

Finally, geopolitical risk can turn bad quickly or good quickly. I think it will turn good. Midterms are approaching, and Americans don’t like the current situation. That’s hard, but I suspect Trump tends to choose de-escalation rather than escalation.

Austin Campbell: That’s it for our time. Thank you so much for joining us today. The timing is perfect, and I hope the audience benefits from your perspectives.