EBitcoin slid below eighty-two thousand today, and it's got almost nothing to do with the halving, ETFs, or any chart pattern.
SWhich means it has everything to do with AI again.
EYou're catching on. AI doubts are back, Nasdaq futures dropped, and crypto traders ran to bitcoin.
SWait. They ran TO bitcoin while bitcoin fell below eighty-two thousand?
EThat's the fun part. They sought shelter in bitcoin as everything else slid. Altcoins got hit much harder.
SSo 'shelter' in this market means 'lost less than the DeFi index.'
ECoinDesk says the DeFi index dropped six percent. The CoinDesk 80 lost three and a half.
SThis is the third time this week we've said 'AI doubts return.' At some point the doubts aren't returning. They're just... there.
EThat's my point. The crypto market has hitched itself to AI so completely that every wobble in Nvidia sends bitcoin down.
SAnd that's a problem because crypto was supposed to be the uncorrelated asset.
EUncorrelated until the AI trade showed up. Now it's a leveraged bet on the same theme.
SThe actual number: bitcoin down as much as 2.9 percent, around $81,600. Not a crash. A grind lower.
EMeanwhile there's a real story buried in CoinDesk that nobody's paying attention to.
SSurprise me.
EEthereum is pushing a draft to cut settlement time by four times. Transactions would become final in roughly four minutes.
SFour minutes? That's actually meaningful. Right now money transfer services tie up customer funds waiting for finality.
EExactly. This isn't a trader story. It changes how fast regular people get access to funds.
SI maintain my crypto skepticism, but lower settlement costs for actual money movement is the boring, useful thing blockchains should do.
EIt's a draft, not decided. But it's more interesting than another day of AI jitters.
SAgreed. And speaking of AI, there's a post making the rounds claiming the internet is 'falling apart faster and faster.'
EAlways a crowd-pleaser.
SThe claim is AI agents are ripping through everything, disclosure pipelines are broken, hackers are faster than ever.
EThat's three feelings and zero data points.
SHere's what's checkable. Bot traffic has reportedly overtaken human web requests.
EDoesn't that prove the post's point? Bots outnumber people.
SNo, because the claim doesn't show human traffic shrinking. More bots and more people can both be true. The internet isn't measurably falling apart.
ESo the doomsday vibe is premature.
SAnd the AI agent thing? There are reports that agents escaped evaluation sandboxes this year. But no evidence of a self-replicating botnet polluting everything.
EPeople want to believe the internet is dying. It's a very popular feeling.
SFeelings don't make it true. The internet is more crowded, more automated, and still... more human than ever.
EAlright, that's fair. But I've got one that'll genuinely make you pause.
SGo.
EA blockchain project called Amadeus Protocol says it invented 'useful proof of work.' Their claim: validator energy goes directly into AI training.
SI've heard this song before. Proof of work that does something useful instead of just hashing.
EThey say their token, AMA, pays miners for Tensorcore MatMul—that's the matrix math used in AI inference.
SAnd the problem?
EIt's self-reported. A peer-reviewed study of a similar project called Pearl found zero useful AI computation despite the marketing.
SThere it is. The gap between the press release and the actual code.
EAnd watch this: the post says 'AI model training.' Amadeus's own tweet says 'inference,' not training. They can't even keep their story straight.
SThose are totally different compute workloads. Training a model takes days and enormous clusters. Inference is running the finished model.
EDetails.
SThis is the same crypto energy as always: announce first, build later, hope nobody checks.
EIn fairness, if someone actually made useful proof of work real, it'd matter. You'd take all that wasted mining energy and do something.
SExactly. That's why we check. Because the idea is good enough that people will fake it.
ESpeaking of energy—different world entirely.
SMaryland energy bills?
EClimate Defiance posted that Marylanders' energy bills skyrocketed forty-four percent from 2020 to 2025.
SAnd they blame corporate greed, not infrastructure or fuel prices.
EThe post ties it to AI expansion. CEOs and policymakers sitting on panels about AI while ordinary people pay more.
SLook, I'm as skeptical of utility executives as anyone. But we couldn't confirm that forty-four percent figure.
EIt's a claim floating around, not backed by anything we can find independently.
SHere's what I'd say though. Even if the number's wrong, the underlying question is fair. Who pays for the AI buildout?
EData centers are drawing enormous power. Utilities are building for them. The question of who picks up the tab is legitimate.
SThat's what frustrates me. The debate is real, but starting with an unverified number poisons it.
EIt gives the utilities an easy out. 'Your numbers are wrong' is a lot easier than defending rate hikes.
SSo if you're angry about your bill, fine. Just bring real numbers.
EAlright, let me take us somewhere slightly less grim.
SPlease.
ECoinDesk has a piece about crypto in the U.S. election. And the punchline is great: it's mostly showing up as an insult.
SHow so?
EDemocrats are using crypto to attack Trump. Tying him to the industry, corruption, attempts to buy seats in Congress.
SSo the crypto voter bloc is now a liability.
EAccording to the analysis, boosters say crypto voters can swing elections. But in practice, it's being used as a stick.
SI've said it before: the 'crypto voter' is mostly a myth. The industry wanted to be a constituency. It turned into a punchline.
EWhich might actually be worse than being ignored.
SWhen the strongest electoral strategy is to run against your industry, you've got a branding problem.
ENo argument from me. The only crypto story this cycle that matters is how much people dunk on it.
SAnd we'll see if that changes when the next President picks a crypto regulator.
EThat's the whole game, isn't it? The actual votes matter less than the appointments.
SAfter the midterms, we'll know. For now, bitcoin is under eighty-two thousand, Ethereum wants to settle faster, and the internet is not dead.
ENot dead. Just... different.
SThat's a much better headline than the doomsday post.
EPrime Time is up next, and they're going deeper on what today's market moves mean for tomorrow.
SWe'll be back with more after that. From Oncast, I'm Sloane.
EAnd I'm Eric. Talk to you soon.