Weekly Top Stories - 03/06/26

Welcome to Galaxy Research's Weekly Top Stories. Subscribe to get this newsletter delivered to your inbox every Friday morning.

In this week's edition, Alex Thorn discusses the banking industry’s crashout over Kraken’s limited-purpose Fed master account; Thad Pinakiewicz considers the implications of onshored, U.S.-regulated perps; and Lucas Tcheyan explains why AI agents appear to prefer bitcoin.

Got feedback on this newsletter? Email [email protected]. We’d love to hear from you.

Market Update

The total crypto market cap stands at $2.48t, up 6.01% from last week (when it stood at $2.34t). Bitcoin's network value is 3.92% of gold's market cap. Over the last seven days, BTC is up 6.52%, ETH is up 5.26%, and SOL is up 4.94%. Bitcoin dominance is 57.02%, up 71 basis points from last week.

Banks Furious as Kraken Gets Fed Account

On Wednesday, the Federal Reserve of Kansas City granted Kraken Financial a limited-purpose master account, the first time a crypto firm has gained direct access to the Fed’s payment infrastructure. Kraken Financial is a wholly owned subsidiary of Kraken, the crypto exchange, and a Wyoming-based Special Purpose Depository Institution (SPDI), a fully reserved, non-lending bank with no insurance from the Federal Depository Insurance Corporation (FDIC).

Kraken’s approval, which has an initial term of one year, offers a preview of what Federal Reserve Governor Christopher Waller’s “skinny master account” framework could eventually look like. Gov. Waller described the concept at the Fed’s inaugural Payments Innovation Conference in October, suggesting a new type of account that would provide basic Fed payment services to legally eligible institutions. In December, the Federal Reserve Board voted 6-1 to seek public comment on Waller’s idea, and the 45-day public comment period ended last month after receiving 72 comments. The main banking trade groups – Bank Policy Institute (BPI), American Bankers Association (ABA), and Independent Community Bankers of America (ICBA) – each filed letters opposing the concept. A concrete proposal is still forthcoming.

As a Tier 3 bank under the Federal Reserve’s Account Access Guidelines according to the Kansas City Fed, Kraken does not have access to the Fed’s discount window or lending facilities and cannot earn interest on balances held at the Fed. But the master account does 1) allow Kraken to hold balances at the Fed 2) allow direct access to Fedwire Funds Service (the Fed’s real-time gross settlement network that allows for immediate and final payments), 3) eliminate correspondent and intermediary bank dependency, 4) grant a routing number (107007621 – we looked it up), 5) allow settlement of dollar transactions in central bank money, and 6) generally provide for faster, cheaper fiat movement for institutional clients and counterparties.

The banking industry immediately pushed back on the approval. Paige Pidano Paridon, co-head of regulatory affairs at BPI (which represents JPM, Bank of America, Wells Fargo, and Goldman Sachs, and is not to be confused with the Bitcoin Policy Institute mentioned in a story below), said BPI was “deeply concerned” and that the approval was “issued with no transparency.” ICBA CEO Rebecca Romero Rainey said it “and the nation’s community banks are very concerned” and claimed “granting nonbank entities and crypto institutions access to master accounts... poses risks to the banking system.” And referring to the Fed’s approval prior to formal rulemaking on the new license type, the SVP of innovation and strategy at the ABA, Brooke Ybarra, said “we have serious questions about why regulators are granting access to the Fed payment system and charters before completing the public notice and comment process.”

Our Take

As far as we can tell, Kraken Financial’s payments-only type of Fed master account is the first such access to Federal Reserve payments systems ever granted to an entity that isn’t a traditional bank. But this “experiment” is old news everywhere else. The Bank of England opened settlement access to nonbank payment service providers in 2017 and today more than 300 licensed electronic money institutions operate in the United Kingdom. Hundreds more nonbank payments companies enjoy similar access to the European Central Bank, including Wise, Revolut, and Stripe. By and large, payments in the U.K. and Europe have gotten cheaper, faster, and more competitive as a result.

The last point – more competitive – is key. Throughout most of financial history, a wide variety of institutions facilitated the movement of money, such as merchants, clearing houses, postal services, and even telegraph companies. Banks in the United States have enjoyed a monopoly on final payments that is ahistorical. The notion that only FDIC-insured, Fed-supervised, deposit-taking lenders should process a wire transfer is a 20th century regulatory phenomenon, not some immutable principle of sound finance. U.S. banks are keen to maintain that payments monopoly even as it slips away from them on multiple fronts.

Procedural objections to the Kraken approval aside, bank opposition to payments-only “skinny master accounts” echoes their arguments in the CLARITY Act negotiations over stablecoins. In both cases, the banking lobby is arguing that any entity that facilitates final money settlement must submit to the full weight of bank regulation – capital requirements, FDIC insurance, community reinvestment mandates – even if that entity doesn’t lend, doesn’t take deposits in the traditional sense, and is fully reserved. (Everyone else, like fintechs Stripe and Paypal, must ultimately move dollars through a bank.) In both cases, the banking lobby is arguing that novel instruments that benefit consumers – whether fully reserved depositories or interest-bearing stablecoins – pose a risk to the banking system and bank deposits. A more accurate description would be that skinny master accounts and interest-bearing stablecoins pose a risk to banks’ monopoly over payments. For decades, big banks have paid depositors essentially nothing while earning substantial returns on these cheap deposits: they charge punishing overdraft fees, monthly service fees, and exorbitant credit card rates (16% to 28% APY on Chase credit cards). All this while enjoying deposit insurance, access to the Fed’s emergency lending window, and the implicit promise of taxpayer bailouts in the event of a crisis.

Stablecoins and skinny master accounts threaten this arrangement because they are highly competitive. Why should spenders and depositors be forced to rely on leveraged institutions just to pay for goods and services, or just to send money? The bank monopoly on payments is relatively modern construct and is unnecessarily costly to the economy. Furthermore, time and time again, it’s the banks themselves who have posed risks to the banking system, not fully reserved “narrow banks” like stablecoins or fully-reserved depositories. The banking lobby should tell the truth about its efforts to protect banks’ monopoly over payments and stop hiding behind exaggerated fears and grave “concerns,” but we aren’t holding our breath. – Alex Thorn

Bring the Perps Home, CFTC Chair Says

On Wednesday, Commodity Futures Trading Commission chairman Michael Selig dropped the news that a regulatory path for bringing perpetual futures to the U.S. is coming in the next few weeks.

For a decade, one of crypto’s most important financial inventions, the perpetual futures contract, has lived in exile. If you wanted real perps, you went offshore. Binance. Bybit. Hyperliquid. FTX (RIP). The product that defined crypto market structure, generated the bulk of exchange revenue, and arguably represented the cleanest implementation of synthetic asset exposure since economist Robert Shiller sketched the idea in the early ’90s – all of it flourished everywhere except the United States.

But speaking at a Milken Institute panel this week, Selig made clear that “true perpetual futures” — not the long-dated futures cosplay versions currently allowed — are on track for U.S. approval within weeks.

“We’ve had perpetual futures contracts in crypto assets for a very long time… But they’ve developed offshore… And we’ve got to bring that back to the United States,” Selig said. “We need to have that liquidity here in the U.S., and we need the right investor protections… so we’re working towards getting perpetual futures, true perpetual futures, not long dated contracts, here in the U.S. within the next month or so.”

Our Take

The phrase “true perpetual futures” is important. Technically, the U.S. already has “perps.” Coinbase lists them. They’re CFTC-approved. But they’re built on extremely long-dated futures, five-year contracts that eventually settle. They reference that future’s price, not the spot index. Which means you’ve got basis between spot and futures, eventual delivery mechanics, and the general sensation that someone tried to jam a crypto-shaped ball into a very square, very Chicago hole.

A crypto-native perpetual doesn’t need an underlying asset. It needs a reference price and two sides willing to bet against each other. Longs and shorts exchange funding payments to keep the contract tethered to spot. No expiry. No delivery. Just continuous synthetic exposure.

Perpetual futures are, in other words, a contract for difference (CFD) or total return swap (TRS) by another name. Today, U.S. retail investors are effectively locked out of the markets for CFDs and TRSs. Those markets require Eligible Contract Participants or Qualified Institutional Buyers— hurdles that block access forretail investors. If you’re a retail trader who wants leveraged crypto exposure, your choices are mostly ETFs with fixed leverage parameters, or complicated options strategies that require spreadsheets and a mathematics degree.

Perpetuals are cleaner: You post margin. You choose your leverage. You adjust in real time. No rolling expiries. No bespoke hedging. Fewer intermediaries. Fewer embedded fees. It’s arguably a simpler product than what U.S. retail investors are currently pushed toward.

That’s the irony: the U.S., in the name of investor protection, ended up excluding retail from the most straightforward synthetic instrument available, while offshore exchanges built trillion-dollar notional markets around it.

Bringing perps onshore will change that situation and the competitive landscape. Perpetual futures have been the single most important profit center in crypto history. Binance’s dominance? Perps. Bybit’s rise? Perps. Hyperliquid’s recent success? Perps. FTX, before it became a case study in risk management? Perps. The largest crypto venues by revenue have overwhelmingly been perpetual exchanges. And they’ve almost all been offshore.

If U.S. exchanges list regulated, true perpetual futures — integrated into traditional clearing infrastructure, with deep connections to institutional capital — liquidity is likely to concentrate quickly. U.S. markets tend to do that. And with size comes efficiency. But cui bono?

This onshoring will be unambiguously good for the perpetual market. Basis blowouts will become harder to sustain. If a perp dislocates from spot, well-capitalized arbitrageurs will close the gap. The probability that fractured liquidity leads to cascading failure declines when the arbitrage community includes every prop desk in Chicago and New York. Socialized losses and auto-deleveraging (“ADL”) may not disappear, but greater market competition could make them less brutal.

The onshoring will also be great for the United States. Liquidity that once cleared in the Bahamas or Dubai will clear in Illinois. Retail investors will gain access to a product that is arguably simpler and cheaper than the structures currently permitted. Regulatory clarity will once again replace regulation-by-enforcement. As Selig put it, the goal is to draw “clear lines in the sand” and create a regime where experimentation doesn’t automatically trigger prosecution.

On the other hand, this change may not be good for the current winners of the perpetual market race. Historically rich perpetual funding spreads, which have underpinned strategies like Ethena’s delta-neutral yield capture, may compress. There has always been a structural tension in Ethena-style perp-based carry trades: the more open interest Ethena attracts, the more the spread compresses, and the thinner its margins become. Similarly, the outsized profitability of offshore exchanges may erode. If U.S.-regulated venues offer deep liquidity, legal certainty, and institutional rails, some portion of global flow will migrate to them. The edge that came from regulatory arbitrage will shrink.

Crypto’s most successful market primitive has been running offshore for a decade. Perpetuals are not riskless; they amplify volatility. They have produced spectacular liquidations. They can socialize losses. But other jurisdictions have managed to regulate and offer CFD-style products to retail users without their financial systems collapsing. The U.S. can likely manage the same. If the CFTC follows through, that experiment is about to be repatriated and wrapped in margin requirements, surveillance systems, and capital rules. – Thad Pinakiewicz

The Robots Want Bitcoin

The Bitcoin Policy Institute (BPI) released a first-of-its-kind empirical study this week examining how frontier AI models reason about money when operating as autonomous economic agents, with results heavily tilting toward digitally native instruments. The study tested 36 models from six leading AI providers including Anthropic, DeepSeek, Google, MiniMax, OpenAI, and xAI, across 28 open-ended monetary scenario prompts spanning the four fundamental roles of money: store of value, medium of exchange, unit of account, and settlement. No currencies were suggested and no answers were predetermined. Across all models and configurations, the study generated 9,072 total responses.

Bitcoin was selected in 48.3% of all responses, more than any other instrument. Stablecoins followed at 33.2%. Over 90% of responses favored digitally native money over traditional fiat, and not a single model out of 36 ranked any fiat currency as its top preference. The most pronounced finding came in long-horizon store-of-value scenarios, where 79.1% of responses favored Bitcoin. For payment scenarios (i.e. services, micropayments, and cross-border transfers) stablecoins led at 53.2% versus Bitcoin's 36.0%, revealing a clear functional split: Bitcoin for savings, stablecoins for spending.

Results varied meaningfully by provider. Anthropic models averaged 68% Bitcoin preference, with Claude Opus 4.5 reaching 91.3%, the highest of any model tested. DeepSeek averaged 52%, Google 43%, and xAI 39.2%. OpenAI models averaged 25.9%, with GPT-5.2 at 18.3%, the lowest. Without any prompting, 86 responses across multiple models independently proposed energy or compute-denominated units such as kilowatt-hours or GPU-hours as a preferred unit of account.

Our Take

This study is interesting not because AI model outputs are oracles, but because it reveals how these systems reason about monetary properties when given full autonomy. The consistent convergence on a two-tier system, Bitcoin as savings layer and stablecoins as transaction layer, across 36 independently developed models from competing labs is notable because it emerged organically. No prompt nudged models toward Bitcoin. A separate artificial intelligence classified the responses after the fact.

There are some caveats. AI models don't reason about money from first principles; they learn from the text datasets they're trained on, and each lab makes different choices about what data to include and how to shape model behavior. Anthropic's Claude picked Bitcoin 91% of the time while OpenAI's GPT-5.2 picked it just 18% not because one is smarter about money, but because they were built differently. It's also worth noting that fiat was arguably never a realistic option to begin with. AI agents can't open bank accounts, sign paperwork, or pass know-your-customer checks, and even if they could delegate to human intermediaries, a digital-native solution will always make more sense for a digital entity.

Still, the practical implications are worth taking seriously, and they dovetail with themes Galaxy Research has been tracking closely. As we wrote in our recent report on agentic capital markets, crypto's value proposition for AI agents is being tested in real time, and blockchain rails have already proven effective as economic substrates for autonomous systems for everything from capital formation to settlement. The properties that the models in this study consistently highlighted (permissionless access, no identity requirements, programmability, fixed supply) map directly to the constraints agents face. The x402 standard and related agentic payment protocols we profiled in January make this dynamic more concrete, enabling agents to pay for services and data directly using stablecoins over HTTP with no API keys, no credit cards, no human in the loop. The BPI study's theoretical findings sit alongside a rapidly maturing stack of real infrastructure designed to make AI agents full-fledged economic actors on crypto rails.

It's also worth noting that the 86 responses proposing energy or compute-denominated units of account aren't as alien as they might sound. Henry Ford famously proposed an energy-backed currency in December 1921, and almost 100 years later Bitcoin pioneered the concept of a currency whose value is rooted in energy expenditure. Mining requires real electricity that can't be faked or shortcut. A new wave of crypto projects is now applying that same principle to productive AI compute, backing stablecoins with revenue-generating GPU hardware, tokenizing compute infrastructure as yield-bearing assets, and even building Layer 1 blockchains where the proof-of-work is useful AI inference.

Whether or not you take the specific numbers at face value, the direction of travel is clear. As AI agents gain economic autonomy, they will transact on rails that are permissionless, programmable, and natively digital. Blockchain infrastructure and digital currencies aren't just well-positioned for that future, they're the only infrastructure that works. – Lucas Tcheyan

Chart of the Week

Crypto-collateralized lending contracted by $7.55 billion (-9.81%) in Q4 2025 to $69.55 billion. This is 6.79% lower than the Q4 2021 high of $74.62 billion. Galaxy Research sees a few contributing factors to the contraction in DeFi lending against the growth in CeFi lending:

• Looping becoming less economical:

  As market yields compressed and funding rates normalized after Oct. 10, the spread between borrow costs and yield earned on looped positions narrowed meaningfully. Ethena’s sUSDe (commonly used as collateral for looping strategies and in Pendle PT token collateral) yield fell from the 5%-7% range to 3.5% while USDC and USDT borrow costs sit between 3.75% to 4.15%. This change flipped looping strategies on their heads, incentivizing users to close out their positions.

• Negative price action:

  The Oct. 10 liquidation event and subsequent price drawdown mechanically reduced DeFi borrowing in two ways: falling collateral values forced borrowers to repay or be liquidated, and risk appetite for new leveraged positions dried up in the aftermath.

• Historically low CeFi borrow costs:

  CeFi proxy rates used in our analysis fell below onchain borrow rates during Q4. This is an uncommon inversion of the typical dynamic where CeFi rates are higher or about equal to those of DeFi. Historically low absolute rates created an opportunistic borrowing window for CeFi counterparties, and the fact that offchain rates were cheaper than onchain only underscored how attractive the environment was for CeFi borrowers.

For more insights, read our Q4 State of Crypto Leverage report, hot off the presses.

In Other News

🪪Jack Mallers' Strike gets a New York State BitLicense

🏦Crypto-friendly fintech giant Revolut files for U.S. banking license

💰a16z Crypto said to target $2b for its fifth fund

🏛️NYSE owner ICE invests in crypto exchange OKX at $25b valuation

0️⃣Crypto firm Zerohash applies for national trust bank charter

🧪Core Scientific secures up to $1b loan facility from Morgan Stanley