# stablecoins — X 热门讨论 (2026-09-23 17:11 UTC)
## @Eliteonchain (Elite🏝) · 09-23 15:17 · ♥62 ↻0 💬27 Almost $1B more in stablecoins is sitting on Hyperliquid than a week ago.
DefiLlama tracks $7.796B on @HyperliquidX’s L1, up 13.56% in seven days. That works out to roughly $931M in additional balances.
The breakdown is even more striking:
> USDC: 99.26% > All other stablecoins combined: 0.74%
So roughly $7.74B of that balance is USDC. For all the competition between stablecoin issuers, this particular ecosystem runs almost entirely on Circle’s dollar.
We usually talk about Hyperliquid through trading volume. But volume measures how much turns over, and the same capital can trade repeatedly.
Stablecoin balances give us another view: how many dollars are actually sitting inside the ecosystem.
Now, that extra $931M doesn’t mean traders opened $931M of new positions. The balance figure alone doesn’t tell us how much is collateral, deployed elsewhere or waiting to be used.
Still, a 13.56% weekly increase is substantial. https://x.com/Eliteonchain/status/2102779486525644964
## @KyleReidhead (Kyle Reidhead | Milk Road) · 09-23 14:30 · ♥46 ↻13 💬9 BlackRock just published a paper saying AI is the most underappreciated demand driver for crypto (holy shit)
This quote was killer:
"AI represents machine-native intelligence, digital assets represent machine-native money"
These two themes have been treated as separate trades
With the famous tweet: "If you're in crypto, pivot to AI"
But I've long said that a bet on crypto is a bet on AI and that's starting to play out
Blackrock notes 2 specific areas where these technologies converge:
1. Agentic Commerce requires machine-native payment rails
As AI agents start actually DOING things (buying data, booking travel, renting compute), they need a way to pay. And that's where it gets interesting
An AI agent can't easily walk into a bank and open an account. Card rails need human onboarding, charge merchant fees that make a sub-penny API call pointless, and take days to fully settle and clear disputes
A stablecoin wallet runs 24/7, settles in seconds, and doesn't care if the owner is a person or a piece of software
2. Compute is emerging as a new and potentially large market for digital assets
As compute becomes one of the largest in-demand products in the world, BlackRock thinks standardized, tokenized claims on compute become a real digital asset market, with agents shopping for GPUs and paying per job.
There are already early signs of Compute markets and digital assets converging with @USDai_Official, who uses stablecoins to provide financing for GPUs and uses the interest to give yield to USDAI stablecoin holders
As the compute market expands, this market should have significant growth across the digital assets ecosystem
Crypto has long been judged/valued based on human use cases, but I think it needs to be looked at as financial technology for AI, rather than humans (though it will work for both)
Companies like Coinbase and Circle are already going all in on becoming the payment structure for AI and so to are many DeFi protocols. If you want to see what assets I hold in my portfolio to capture the agentic financial upside, you can check out my portfolio here: https://t.co/NyRpAKYOwu
Don't forget to give me a follow @kylereidhead for more insights on AI, crypto and markets > 引用 @milkroaddaily: BlackRock's new report says AI and Crypto are complementary technologies.
They share three zones of overlap:
> Shared tokenization architecture: LLMs break language into tokens machines can understand. Blockchains break economic value into tokens machines can verify, transfer, and settle.
> Agentic commerce needs machine rails: Cards, ACH, and bank accounts were built for humans clicking buttons. Agents need 24/7, programmable, high-frequency rails, which crypto-powered protocols like x402/MPP provide.
Compute is becoming a digital asset market: Hyperscaler cloud revenue is projected to reach ~$1.1T by 2030. Standardized, tokenized claims on compute capacity could be financed, hedged, and settled onchain.
Citing simulations by the Bitcoin Policy Institute, BlackRock further said that ~79% of models chose BTC for value storage, while 52% chose stablecoins for transactions, with bank money at under 9%.
If the thesis plays out, crypto isn't just a side hobby for AI; it becomes the structural buyer of:
> Settlement and blockspace > Stablecoin float and velocity > Programmable assets (tokenized cash, treasuries, RWAs) > Eventually even compute claims
Coinbase and Solana are already leading in x402 agentic payment transaction volumes, while Stripe-backed Tempo is lining up agentic transactions with its MPP protocol.
Data from Artemis already show agentic transactions going from virtually zero in Q4 2025 to ~262M till date.
Circle-backed USDC is the workhorse asset for x402-style flows. Circle's recent launch of the ARC Network, with the intention of becoming the settlement layer for agentic transactions, proves they're already building products to lead this category.
BTC, ETH and SOL are all big beneficiaries of the machine economy.
BTC is the savings asset for AI. ETH's default settlement-layer vibes, combined with its lead in stablecoins and RWAs, mean more and more agents will use the network for identity, wallets, and transactions. SOL, on the other hand, is the best positioned for low-cost, high-volume transactions. https://x.com/KyleReidhead/status/2102767459023622277
## @tehMoonwalkeR (tehMoonwalkeR) · 09-23 14:29 · ♥51 ↻5 💬6 I cannot get over the fact that @BlackRock confirmed the $SUI thesis !
$SUI has been literally building their chain with this narrative in mind
Now Franklin Templeton and BlackRock, ~$12 trillion combined, have published the same thesis independently, months apart 🧠
📖 BlackRock: AI and crypto share the same foundation. LLMs tokenize language, blockchains tokenize value → That's Sui's object model, described from the outside
💳 BlackRock: agents need machine-native rails, banks are too slow and too expensive for micropayments → @SuiNetwork killed its own fee revenue in May specifically to win this
⚙️ BlackRock's own example needs FOUR stitched-together protocols for one agent payment → A single Sui PTB already does it all, atomically, in one transaction
💰 Their numbers: stablecoins at $300B+ circulating, $11T+ transacted in 2025, growing 80% CAGR vs ACH's 8.5% → Exactly the volume gasless transfers and USDsui were built to capture
🖥️ BlackRock flags compute as the next tradeable asset class for agents → The one frontier still open. Nautilus & @WalrusProtocol is early infra here, the rest is unclaimed https://x.com/tehMoonwalkeR/status/2102767297773371610
## @martypartymusic (MartyParty) · 09-23 16:12 · ♥53 ↻3 💬7 Tether declined to apply for a MiCA license in the EU over reserve rules.
Tether CEO Paolo Ardoino said the rules require 60% of significant stablecoin issuers’ reserves in commercial bank deposits, making stablecoins less protected. EU deposit insurance covers only €100,000.
“When MiCA becomes safer for consumers and stablecoin issuers, we may reconsider the decision,” Ardoino said.
IMO: I agree - forcing stable coin issuers to back with bank liquidity is too risky going into 2030. Bank ledgers are too high a risk. If a EU Bank collapses the Stable Coin are unbacked. Soverign debt is acceptable but bank debt is not. https://x.com/martypartymusic/status/2102793300860109269
## @demian_ai (dylan ツ) · 09-23 15:18 · ♥40 ↻1 💬5 boy this is fascinating 👀
-> what if the buyers of compute become agents?
i read blackrock’s new AI paper so you don’t have to (it's secretly an inference paper)
i like how they frame this idea: 1. AI as machine native intelligence 2. digital assets as machine native money 3. agents as the glue
the tell they lean on is Stripe recently agreeing to buy Openrouter. Openrouter sits in front of 400+ models across 80+ providers (including @nebiustf) and routes by cost and fit. Stripe handles the payments. So easy to put them together and thinking that shopping for tokens could become billing infrastructure.
but let's talk about the compute part
Capex still gets the headlines. Some estimates they cite put AI build north of $5T through 2030. They have another point: watch opex. The hyperscalers are on a consensus path toward $1.1T rev by 2030. Right now training is concentrated to a few buyers, long contracts, with humans on both sides of the table.
inference is the mass market. The mckinsey chart they reprint is the whole essay in one figure: by 2030 inference is the largest slice of AI data center power, something like 43% in their cut.
-> It won 't only be chatbots, it will be millions of buyers asking for a unit of useful work: tokens, a job, reserved capacity, a region, a latency, a price.
so quite an interesting power-market logic when u think about it.
When the buyer is an agent, a normal card payment starts to feel weird (e.g the ticket might be tiny).
Blackrock points at always-on rails (stablecoins, x402-style "pay before the API answers" type) as the plumbing that might fit.
once that works, inference starts looking like something you can shop and settle without a person in the loop.
machine native intelligence needs machine native money. The quiet third term is machine native inference? > 引用 @BlackRock: Our latest research paper explores the growing connection between AI and digital assets and explains why broad AI adoption may drive new demand, utility and applications across the digital asset economy. https://t.co/z5T88Orble https://t.co/GMxQqTVmUN https://x.com/demian_ai/status/2102779672496865305