Hyperliquid Strategies reported $620.6M in net income for the third quarter.
Total quarterly revenue landed at $13.3M. The massive profit swing was driven by mark-to-market gains on its HYPE holdings.
The NASDAQ-listed entity holds substantial token balances and collects validator staking rewards.
Traditional public companies rely on operating margins to justify valuations.
Hyperliquid Strategies is acting like a leveraged corporate treasury proxy for the L1.
When token market caps swell, corporate balance sheets look like hedge funds.
Bitcoin Core developer Cory Fields proposed a concept called quantum landmines on Thursday.
The framework creates an economic deterrent against future quantum computing attacks.
Quantum computers could theoretically steal funds from older addresses with exposed public keys.
Fields proposes requiring anyone moving legacy coins to put up matching Bitcoin collateral.
A waiting period allows legitimate owners to reveal proof and seize the attacker's collateral.
An attacker would risk losing their own capital on every single attempted theft.
Economic game theory is being used to protect Bitcoin before quantum hardware exists.
Hyperliquid Strategies has filed with the CFTC to become a Futures Commission Merchant.
The application was submitted through its operating subsidiary, HypeStrat Markets.
The goal is to legally handle US client accounts for permissioned HIP-3 markets.
HIP-3 allows third-party builders to launch perpetual markets on Hyperliquid infrastructure.
Hyperliquid's core offshore exchange still blocks direct access to US IP addresses.
The company is trying to build a regulated onshore bridge for institutional capital.
Bringing perpetual futures inside US regulatory boundaries is the real battleground.
fwiw the 5 hour removal was only for verified ultra accounts, newer ones keep it until anthropic verifies them
might be why it came back for you
It drives me insane when people say Claude performs better at Medium effort than at High.
Extra is still my daily driver, and Max is my favorite.
Max has always been too slow, but it produces the best visual outputs.
I tried dropping down to High, but it’s just not on the same level as Extra. I only use Medium for simple tasks.
That said, Opus 5.5 is becoming less token-efficient, and now I hit my usage limits way quicker.
On top of that they brought back the 5-hour limit just a few days after removing it, without saying a word. No comment.
Horizen's Layer 3 network on Base was forced to shut down during the Denim fork.
The L3 settlement architecture could not process Base's 10x increase in block velocity. Teams operated emergency bridging procedures to move user capital back to L1.
Layer 3 networks settling on L2s rely entirely on predictable parent block cadences.
When an underlying rollup accelerates its clock, downstream modular layers break.
Modular stacks promise infinite horizontal scale until parent upgrades break compatibility.
Building rollups on top of rollups multiplies dependency risks instead of reducing them.
Bill Gates on AI-enabled bio risk: "I'm hopeful we get lucky on that one"
"Bio is, you could say, even scarier because the scale of the damage, you could literally have something that is significantly worse than Covid."
"Now bio, there's not a business model for. So actually predicting when somebody would do that is more difficult."
"There's a lot of things having to do with the safeguards in the models and a lot of surveillance so we could catch something very quickly. There's a lot we need to do that we're just not doing."
Base's Denim upgrade introduced a critical trap for legacy smart contracts.
The network introduces BaseTime to deliver millisecond-level Unix timestamps on-chain.
However, block numbers now advance ten times faster in real-world time. Any contract using block counts to calculate duration will shrink by 90%.
Vesting schedules, governance voting windows, and timelocks will expire prematurely.
Developers who failed to audit their contract logic face unintended state changes.
Speeding up block frequency creates instant friction for immutable protocols.
at least the third trillion dollar wipeout this year, and each one got called the end of ai
the capex kept coming after the other two
agree on payments though, they hold up in drawdowns
$2.9 trillion was just wiped off the U.S Stock Market.
the AI Bubble has officially popped.
the days of investors & VCs blindly throwing cash at anything AI related has come to an end.
ironically, payments and fintech companies tend to be the most resilient businesses during market downturns like these.
i predict we’re in store for an upcoming fintech super cycle.
Base deployed its Denim upgrade to mainnet on Thursday at 18:00 UTC.
Canonical block times dropped from two seconds to 200 milliseconds. The network now produces five complete canonical blocks every single second.
Coinbase's L2 officially retired Flashblocks because sub-second blocks are now native.
RPC providers and decentralized applications had to reconfigure data streaming pipelines.
Sub-second settlement closes the speed gap between rollups and centralized order books.
Base is turning Ethereum L2 execution into high-frequency trading infrastructure.
🇺🇸 Patronus AI has open-sourced GlobeBench, a set of 1,632 tasks that test whether LLMs can simulate RL environments
Each task asks a model to predict what the environment returns next
GPT-6 Astra gets 49.6% of them right
Grok 4.6 is second at 28.1%
Claude Opus 5.5 is third at 26.4%
None of the other 19 models tested reaches 30%
Bill Gates says AI regulation is stuck because the US doesn't even have mandatory review of new AI models
"Each company, when they were created, they had a notion that this technology was different. OpenAI had a clause that said if it passed AGI, they could ignore the capitalistic tendency to put it out there and just not ever release it. The DeepMind founders negotiated with Google to have some special treatment."
"But as soon as we had multiple companies, then the only way to solve this moves up to the government, you know, it has to decide is open weights okay. And when you have a new model, do you have a benchmark about the monitoring and the safeguards that are in that model?"
"So far the US doesn't even have mandatory review, and so other governments, including China, you know, there's not much value in acting unless the US is part of that. So we're kind of [stuck]."
"I think the US will come along on this. The only question is how many bad events need to take place before they change their posture."
Spot Bitcoin ETFs suffered $430.2M in net outflows on Thursday.
Wednesday's small $74.6M inflow proved to be a single-day pause.
Fidelity FBTC led the selling with $199.9M in net redemptions. BlackRock IBIT followed closely with $105M leaving the fund.
Grayscale lost another $49.9M, while VanEck saw $32.9M exit.
Every major issuer recorded either outflows or zero new net subscriptions.
Institutional buyers are hitting the sell button the moment macro volatility picks up.
🇺🇸 Artificial Analysis found that nearly 1 in 4 of the 100+ serverless endpoints it tested score below the reference
Some providers quantize weights or cap output lengths to save on speed and cost
Two GLM-5.3 Flash endpoints capped at about 32K output tokens scored 22% and 29%
The reference scored 55% on the same test
The cap cut off roughly half of their answers before they finished
Over $1B in leveraged crypto derivatives was wiped out in twenty-four hours.
Bitcoin dropped from $80,500 to $75,500 in a steep afternoon cascade.
Ethereum shed more than 7%, breaking below $2,250. Major exchanges liquidated more than 211,000 traders.
Wall Street tech equities sold off as the VIX volatility index spiked.
That traditional market risk-off took crypto leverage down with it.
Traders expecting a clean post-election continuation were completely caught offside.
Monday’s seven tripwires, scored through Thursday.
Four tripped.
Binance net outflows peaked near 3% of platform assets.
The Nasdaq fell 5.52% on Thursday as chips sold off.
Bitcoin hit $75,463 after peaking at $81,105 on Wednesday.
Binance’s proof of reserves cleared.
The jobs report lands today at 13:30 UTC.
Brent and Iran stay open.
🇺🇸 Inkling 2 Small from @thinkymachines is the cheapest model on Surge AI’s sudo L7 staff-level software engineering benchmark
It scores 14.1 at $0.34 per task
Gemini 4 Argon scores only two points higher at $2.64 per task
Muse Wonder 1.0 sits within two points as well, at $4.39 per task
Muse Spark 1.3, Gemini 3.8 Flash and GLM 5.3 all score lower while costing between 3.4x and 7.3x more
🇺🇸 FrontierPhysics is a new benchmark that tests agents on end-to-end physics research
It covers 56 tasks across 9 domains, built by 42 physicists from 26 institutions
Claude Opus 5.5 leads with an overall score of 27.4
GPT-6 Astra comes second at 23.2
Opus scores 45.8 on the verifier, well above its overall score
Gemini 3.8 Flash and Mistral Large 4 both score 0.0 overall





















