Pinned Tweet
Hyperliquid may have just changed how people think about IPO price discovery.
The IPO priced at $135, but SPCX-USDC on Hyperliquid never really traded like $135 was the fair value.
As IPO details came out, SPCX-USDC sold off under $200, then spent days prior to the IPO chopping in the $150-$180 range.
On IPO day $SPCX traded much closer to that range than the official IPO price.
Most people without an allocation were never getting filled anywhere close to $135, and the first real fills were closer to where Hyperliquid had already been pricing it.
The obvious value of tokenized pre-IPO shares is access; but the deeper value is price discovery, and Hyperliquid just proved it can be a real pre-IPO price discovery engine.
Hyperliquid showed that a synthetic pre-IPO market can become a better price discovery venue than the actual IPO process itself, and I think this really changes the way people think about these markets.
The access matters as it allows retail to trade private companies, because defi can create a more efficient public market around tradfi assets before tradfi lets most people participate
If @HyperliquidX can price SpaceX before Nasdaq opens, then future IPOs may have two prices: the official allocation price, and actual fair value the market discovers first.
Anthropic is the next obvious test for this.
so the binance hacker might get caught because one transaction went through metamask's servers instead of a regular rpc
now think about how many apps send every user's transactions through someone else's infra
that's why we keep building @Trustware_io to rely less on third parties
A potential slip by the Binance attacker? @MetaMask should preserve whatever web2 logs they can.
The attacker sent 175M USDT to a fresh wallet.
About 2.5 minutes later, that wallet also received 2 ETH. Likely for gas, matching the pattern used with their other attack wallets.
Exactly two minutes after that, it used MetaMask's Gas Station. Despite already having gas.
Three consecutive transactions landed in the same block: an ETH top-up from Gas Station, an unlimited USDT approval for @KyberNetwork's router, and a payment of 127.67 USDT to Gas Station.
The ETH top up exactly matched the combined maximum gas budgets of the Kyber approval and the USDT payment. Down to the wei.
But they already had more than enough ETH to cover both.
My theory: they were using MetaMask Wallet and, while rushing, selected USDT for gas by mistake
MetaMask sometimes defaults like this for convenience. And the same block top-up sequence matches how their implementation works.
To use Gas Station in MetaMask you have to use MetaMask Sentinel. Meaning the transactions go through them, not a regular RPC.
That interaction probablly left offchain records with MetaMask, potentially including IP metadata.
Among all the attacker wallets I checked, this was the only one showing Gas Station usage.
The wallet: 0xAa246BF639586a6Ba46762D378CE50cA4fc654a0
your LP position is probably asleep right now, parked in one pool on one chain earning on a fraction of the volume it could be
@AquaZero0 lets one position back every pool it's committed to, across uniswap v4 hooks, 1inch aqua, orca and raydium
on aqua0 the same balance can back openai and anthropic pairs on @orca_so and @Raydium at the same time and earn fees on both, which i think is one of the clearest examples of capital efficiency in defi right now
the other half of the flow is getting capital in, which is where @Trustware_io comes in:
you start from whatever you already hold on whatever chain, and trustware routes, converts and settles it straight into your aqua0 vault as a credited balance, non-custodial the whole way through with no manual bridging or swapping first
season 0 is live now and it rewards liquidity that actually gets used, not deposits sitting idle, so if you've got capital parked somewhere doing nothing this is a good time to put it to work with aqua0
a good majority of the issues or bugs we hit testing @Trustware_io were on the rpc side
so we continue to move towards being less reliant on third party providers
partly for performance, which is crucial for Trustware
partly for the non-custodial ethos we've always pushed forward with
i think the teams that actually own their infra are the ones whose execution will actually hold up
we keep building
FullyPorted retweeted
fellas, if your girl:
- is always there when you need her
- never asks you to choose
- makes everyone around you better
that's not your girl. that's @AquaZero0
thank you sk hynix 🫶🇰🇷
what bubble?
SK HYNIX EARNINGS JUST CAME OUT
*SK HYNIX 3Q OPER PROFIT 83.33T WON, EST. 77.5T WON ( BEAT ✅ )
*SK HYNIX 3Q SALES 104.62T WON, EST. 98.96T WON ( BEAT ✅ )
$SKHY
the most underrated part of @Trustware_io : we resolve the route and prepare the transactions but the app keeps its own signing infra the whole way through, so at no point is a third party sitting between your users and their funds
if you're building a protocol with complex onchain actions (multi-step swaps, bridging, contract calls that depend on your own logic) you really can't afford to hand that flow to a custodial provider; you inherit their downtime, their compliance decisions and their risk, and your product only works as long as they let it
with trustware the application actually owns the infra, our paymasters for example are deployed and owned by the client, not by us, so you get the routing and execution without ever giving up control
i think non-custodial infra is going to be the default for any team serious about building in defi, and we're building trustware so teams can do anything and everything onchain without trusting a middleman to do it 🤝
onchain traders already have a price on anthropic and it's weeks before any bank sets the ipo range
@tradexyz listed ANT on @HyperliquidX at $125; on day one it opened at $138, fell as low as ~$109 and closed around $114
i think this is one of the more interesting things defi is doing for retail. getting exposure to private companies has mostly meant secondary platforms with accreditation rules and high minimums, and now anyone can speculate while x reads the s-1 and argues about revenue, compute commitments and whether AI is a bubble
yes, you aren't actually buying anthropic stock; ANT is cash settled with no shares, shareholder rights or ipo allocation, and it runs on the contract's own pricing rules
the s-1 gave wall street the numbers but onchain traders got to price them first
and if ANT ends up anywhere near where anthropic actually lists i think every big ipo is going to do volume on hyperliquid pre-markets from now on
coreweave cds are still trading near record levels and everyone is calling the top, but anthropic's s-1 says otherwise
- $11.6b in q2 2026 revenue, up ~1,370% yoy from $787m
- first gaap operating profit in any quarter $974m (8% margin)
- $3.9b in operating cash flow for H1
demand is showing up and I don't think this looks like a bubble
the AI bubble scare grew in parallel with the era of AI slop (merriam-webster literally made slop the 2025 word of the year), because everyone sees it all day long
we see so many articles and demos; on x we're used to it and can spot slop instantly, but now every social media is filled with it and everyone is seeing it
~48% of anthropic's h1 revenue came through cloud marketplaces like aws, google cloud and azure
now every major lab is making a massive push for retail, building consumer agents to rival what @openclaw and @NousResearch built
muse billboards from meta, openai launched dots, anthropic launched claude duet, and grok ugc ads are all over instagram (@xai is actually bumrushing instagram ads and it's kind of funny ngl)
on the compute side: anthropic has hundreds of billions in compute commitments, and some of those depend on neoclouds lining up money and sites (nscale "best efforts" to get financing, and lambda's $34.9b deal can be terminated if lambda doesn't secure its data center lease), so the coreweave credit scare might send more shocks through AI's supply chain than it will for demand
98% of us households do not pay for AI yet; will it be a bubble if the labs can win retail the way they've won enterprise?
coreweave cds are still trading near record levels and everyone is calling the top, but anthropic's s-1 says otherwise
- $11.6b in q2 2026 revenue, up ~1,370% yoy from $787m
- first gaap operating profit in any quarter $974m (8% margin)
- $3.9b in operating cash flow for H1
demand is showing up and I don't think this looks like a bubble
the AI bubble scare grew in parallel with the era of AI slop (merriam-webster literally made slop the 2025 word of the year), because everyone sees it all day long
we see so many articles and demos; on x we're used to it and can spot slop instantly, but now every social media is filled with it and everyone is seeing it
~48% of anthropic's h1 revenue came through cloud marketplaces like aws, google cloud and azure
now every major lab is making a massive push for retail, building consumer agents to rival what @openclaw and @NousResearch built
muse billboards from meta, openai launched dots, anthropic launched claude duet, and grok ugc ads are all over instagram (@xai is actually bumrushing instagram ads and it's kind of funny ngl)
on the compute side: anthropic has hundreds of billions in compute commitments, and some of those depend on neoclouds lining up money and sites (nscale "best efforts" to get financing, and lambda's $34.9b deal can be terminated if lambda doesn't secure its data center lease), so the coreweave credit scare might send more shocks through AI's supply chain than it will for demand
98% of us households do not pay for AI yet; will it be a bubble if the labs can win retail the way they've won enterprise?
i love seeing the tokenized stock tickers all over my feed
defi doomers are seething and coping, tradfi twitter is still confused
and we keep building
Hermes changed my life more than ChatGPT did…
If you disagree, it’s a skill issue
@NousResearch
