Bitcoin spent the weekend slamming its head against $80,000 and bouncing off. We're sitting around $77,500 right now after a flash crash that liquidated nearly $295 million in positions, mostly long. Strategy added another 3,273 BTC at an average of about $78,000, pushing Michael Saylor's stack to 818,334 coins. Spot Bitcoin ETFs are riding their longest inflow streak of the year, $933 million last week alone. Bitcoin Core v31 is out, and it's a big one — cluster mempool plus Tor and I2P broadcasting on by default. Tesla confirmed a $25 billion AI and robotics capex plan, with Optimus mass production starting in Fremont this summer. And the Fed meeting Wednesday plus GDP and PCE Thursday set up a 48-hour window that could reprice everything. Let's get into it.
Tesla just told the world it's not really a car company anymore. The 2026 capex number is over $25 billion, roughly three times what they spent in 2025, and most of it is going into AI, robotics, and in-house chip fabrication with SpaceX. The Fremont plant — the original Model S and X line — is being converted into a humanoid robot factory. Musk is calling Optimus the biggest product Tesla will ever ship.
The numbers are wild. Fremont is targeted for 1 million Optimus units a year. Gigafactory Texas, eventually 10 million. Mass production starts around July or August. Target unit cost at scale: $15,000. Morgan Stanley figures that capturing just 5% of the 2030 industrial robotics market gets Optimus to about $9 billion in annual revenue. Internal trials reportedly show 92% task completion in battery cell assembly, with potential to cut direct labor in repetitive manufacturing by up to 40%.
Whether you believe the timeline or not — and Tesla timelines are Tesla timelines — the capital commitment is real. They're guiding to negative free cash flow for the rest of 2026 to fund this. That's a serious signal.
And then there's the optics side. Tesla parked Optimus at the Boston Marathon finish line last week to cheer on runners and pose for selfies. Cute marketing. But the same week, a Chinese humanoid from Honor — a Huawei spinoff — actually won a half marathon in 50 minutes and 26 seconds. Seven minutes faster than the human world record. Robots and humans in separate lanes, sure, but several humanoids finished under an hour while the top human ran 1:07.
The contrast tells you something. Tesla is selling the vision and the manufacturing scale. Chinese firms are quietly shipping working units that do specific things very well. Fanuc, Yaskawa, Honor, Unitree — they're not waiting around. If Tesla hits even half of its Optimus production targets, it reshapes labor economics in OECD countries. If it slips on rare earth magnets or vision-system silicon, competitors close the gap fast. Worth watching, but skeptically.
Bitcoin Core v31 just shipped, and there are two changes worth understanding even if you're not running a node yourself.
First: cluster mempool. Until now, nodes evaluated transactions one at a time when deciding what to keep, relay, and stuff into a block. The problem is that real Bitcoin usage involves chains of related transactions — child-pays-for-parent, replace-by-fee, Lightning channel operations. Cluster mempool groups related transactions, up to 64 transactions or 101 kilobytes, and evaluates them as a unit. That means better fee estimation for wallets, smarter block construction for miners — they can pick the most profitable cluster instead of getting tripped up by an underpaid parent — and more efficient eviction when blocks are full.
If you're an ordinary user, you won't notice. If you're using Lightning, doing CPFP, or running anything fee-sensitive, this is a quiet but significant upgrade.
Second change: privacy by default. Bitcoin Core will now broadcast your transactions over Tor or I2P when those are available, instead of the clearnet. The sendrawtransaction RPC supports private broadcast. The point is to stop leaking your IP address to the first peer that receives your transaction. For years, this was the easiest way for chain surveillance firms to tie addresses to physical locations. Now it's mitigated at the protocol-software level, on by default.
Mainnet rollout is expected in the second half of 2026 as nodes upgrade.
Separately, MIT's Neha Narula put out a proposal for a staged path to making Bitcoin quantum-safe. The pitch: don't wait for the perfect answer. Deploy a post-quantum-safe output type via soft fork, add a post-quantum signature opcode, and use BIP 360's Pay-to-Multi-Root structure to let users migrate funds into a quantum-resistant address type. Critical caveat — don't reveal old public keys, which means stop reusing addresses.
This is the constructive side of the quantum debate. The destructive side is Paul Sztorc's eCash hard fork proposal, which would split the chain and reassign Satoshi's coins. The community response, predictably, is calling it theft. Sztorc has been pushing Drivechains for years, but reassigning anyone's coins — even dormant ones — is a non-starter for most Bitcoiners. The line between protecting old coins from quantum attack and confiscating them is exactly where this debate is going to live for the next several years.
The price action this weekend was instructive. Bitcoin pushed up toward $80,000 on Friday, partly on Iran-related headlines, then got rejected and slid. Monday morning, Europe opens, BTC dumps to $77,800, and we get $295 million in liquidations in 24 hours. Earlier in the session, $71 million liquidated in a single hour, with $69 million of that from longs.
But here's where it gets interesting. Funding rates have been deeply negative for weeks — the 30-day cumulative funding rate is around -7%. Shorts are paying longs to keep their positions open. Normally that screams bearish sentiment. Except 10x Research is arguing the negative funding isn't bearish positioning at all — it's institutions running structural hedges, basis trades against spot.
Meanwhile, on Hyperliquid, whales — accounts running positions over $10 million — have flipped from net short to their most aggressively long stance ever. These traders have historically led spot moves by days or weeks. Combine that with Binance reserves dropping by 44,000 BTC over the recent stretch, meaning coins moving off exchanges into custody, and the picture is split: derivatives screens look bearish, spot demand and accumulation look strong.
That's a setup that resolves violently in one direction or the other. If spot pushes through $80,000, the negative funding becomes fuel for a short squeeze. If it fails, leveraged longs cascade like we saw this morning.
ETF flows favor the bull case. Nine consecutive days of inflows through April 24, $2.12 billion added since April 14. Crypto ETFs total AUM is back to $155 billion, the highest since February. BlackRock leading. Strategy added 3,273 BTC last week at $78,000 average, total holdings now 818,334 BTC purchased for nearly $62 billion at an average around $75,500. They're inching toward their 1 million BTC target.
The macro window is the wildcard. The Fed concludes Wednesday afternoon. First-quarter GDP drops Thursday morning. PCE inflation right after. That's a 48-hour stretch where any reaction to the Fed gets immediately overwritten by the data. One analyst is flagging $57,000 as a historical-average bottom if things go wrong. Another says the pullback from $80k is temporary. Both can be right depending on what Powell says and what PCE prints.
The GENIUS Act was signed in July 2025. Full implementation is targeted for January 2027. The interesting period is right now, because the actual rules — the ones that decide who gets to issue payment stablecoins in the United States and what they have to do — are being written this spring.
The new instrument is the payment stablecoin. The new licensed entity is the Permitted Payment Stablecoin Issuer, or PPSI. Treasury, the OCC, FDIC, NCUA, FinCEN, and OFAC are all in the kitchen.
On April 10, FinCEN and OFAC dropped a joint proposal extending AML, counter-terrorism financing, and sanctions rules to PPSIs. The approach is technology-neutral — they're not writing crypto-specific rules, they're applying existing bank-style frameworks. PPSIs will need risk-based AML programs, sanctions compliance programs with internal controls, audits, and training, and a US-based compliance officer subject to FinCEN oversight. Civil penalties go up to $100,000 per day for sanctions violations. Comments are due June 9.
The American Bankers Association and three other banking groups are pushing back. Not on substance — on sequencing. The OCC is supposed to be the primary regulator for nonbank stablecoin issuers, and the OCC's framework hasn't been finalized yet. Treasury, FDIC, and the joint FinCEN-OFAC rule are all interdependent with whatever the OCC ends up doing. The banking groups want the comment deadlines extended to 60 days after the OCC finalizes its rule. Otherwise everyone is commenting blind.
This matters because the GENIUS Act doesn't just regulate Tether and Circle. It defines a category. Whoever ends up qualifying as a PPSI gets to participate in what could become a parallel dollar payments rail. Banks want in. Crypto-native firms want in. The licensing pathway, the reserve requirements, and the supervision regime determine who actually competes.
Meanwhile, the private sector isn't waiting. Western Union is launching its USDPT stablecoin in May. DoorDash is integrating stablecoin payouts via Stripe-backed Tempo across more than 40 countries — this isn't a pilot, it's core labor infrastructure for Dashers in markets where bank rails are slow or expensive. South Korea's KBank is testing Ripple's Palisade wallet for cross-border transfers as an alternative to SWIFT. Banking Circle just launched a euro stablecoin settlement product under MiCA.
The pattern: stablecoins are quietly becoming plumbing. The US regulatory framework will determine whether American firms lead that buildout or whether the action moves offshore.
One thing to sit with. The most consequential Bitcoin upgrade in years just shipped quietly — privacy on by default, smarter mempool, no fanfare. Meanwhile the loudest stories are humanoid robots at marathons and forks that try to reassign Satoshi's coins. The signal isn't always where the volume is. If you run a node, upgrade to v31 when you're ready. If you don't, the network just got better for you anyway. That's how Bitcoin is supposed to work.