Bitcoin jumped back above seventy-one thousand dollars on Monday as geopolitical tensions around the Iran conflict began to cool and the dollar weakened. Strategy logged its biggest ever STRC equity issuance day, buying an estimated fourteen hundred and twenty bitcoin in a single transaction. The ECB dropped a working paper warning that dollar-pegged stablecoins could seriously undermine eurozone monetary policy. Ninety-five percent of all bitcoin that will ever exist has now been mined, crossing twenty million coins in circulation and raising fresh questions about long-term network security. And in AI, Microsoft released a compact fifteen-billion parameter multimodal reasoning model while the open-source LLM leaderboard keeps reshuffling with Chinese and European contenders climbing fast. A lot to unpack today.
Let's start with the market. Bitcoin climbed back to seventy-one thousand dollars after a wild stretch driven almost entirely by the Iran conflict. Crude oil spiked to a hundred and twenty dollars a barrel overnight on war fears, then plunged back to around eighty bucks after Trump made comments suggesting U.S. military objectives were quote pretty well complete. That whipsaw in oil took the dollar down with it, and bitcoin caught a tailwind.
What's really interesting here is what happened on the way down. When bitcoin dipped below seventy thousand, Glassnode data showed traders snapped up nearly six hundred thousand BTC over just two weeks. That is enormous accumulation. Strategy was a big part of that, recording its largest single-day equity raise through its STRC instrument, funding roughly fourteen hundred and twenty bitcoin in purchases. They even amended their sales agreement to allow multiple agents to execute outside regular trading hours, which tells you how aggressively they're moving.
U.S. spot bitcoin ETFs added a hundred and sixty-seven million in inflows on Monday while altcoin ETFs continued bleeding. That divergence matters. Money is flowing specifically into bitcoin, not crypto broadly. There are early signs of capital rotation from gold too. Gold ETFs have seen record outflows after their historic rally, and some of that appears to be moving into bitcoin products.
Meanwhile Bhutan quietly sold forty-two and a half million dollars worth of bitcoin this year, drawing its national stack down from roughly thirteen thousand BTC at peak to under fifty-four hundred. That's a fifty-eight percent reduction. Sovereign selling from a small player, but it does raise questions about how nations manage bitcoin reserves when fiscal pressures mount.
On the technical side, there's a liquidity sweep setup forming around seventy-two thousand with thin supply above and heavy clusters below, so traders should be prepared for volatility in both directions. The broader downtrend hasn't definitively reversed, but the demand floor looks remarkably solid.
Now to something that deserves more attention than it's getting. The European Central Bank released a working paper that essentially says stablecoins are a threat to the eurozone's ability to conduct monetary policy. And honestly, the math they lay out is worth taking seriously, even if you're skeptical of central bank framing.
Here's the mechanism they're worried about. As more Europeans move deposits from euro-denominated bank accounts into dollar-pegged stablecoins, banks lose cheap stable funding. They're forced to rely on more expensive wholesale funding with volatile interest rates. That makes it harder for ECB rate changes to flow through to actual lending. Their estimate is that a ten percent increase in stablecoin market cap could reduce bank lending by zero point two percent. That sounds small until you realize stablecoins are now at roughly three hundred billion in total market cap and growing fast.
ECB Vice President Luis de Guindos and President Christine Lagarde are both pushing for stronger regulation, and the digital euro project is framed explicitly as a sovereignty tool. Euronews ran a piece calling the digital euro quote the EU's tool for payment sovereignty, which tells you how Brussels is positioning this. They see American dollar stablecoins as a backdoor dollarization of European finance.
From a Bitcoin maximalist perspective, there's deep irony here. The ECB is worried about losing monetary control to private stablecoins, but the real threat to their system isn't Tether or Circle. It's that people are discovering they can hold a truly neutral, non-sovereign money. The digital euro, if it ever launches, will be programmable government money with all the surveillance and control that implies. Stablecoins are a halfway house. Bitcoin is the destination.
Meanwhile in the U.S., the Treasury made a notable policy shift, stating that lawful crypto users may use mixers for financial privacy on public blockchains. That's a real change in tone. It keeps the money laundering enforcement framework intact but explicitly opens space for privacy tools within regulated markets. And the CFTC chair is backing blockchain-based prediction markets as quote truth machines. The U.S. regulatory picture is becoming more nuanced even as Europe circles the wagons around monetary sovereignty.
One complication in Washington though. Trump's threat to block all Congressional action until he gets a voter ID bill passed has put the broader crypto legislation agenda on shakier ground. These things are interconnected and the stablecoin bill could be collateral damage of unrelated political fights.
Let's talk AI. The open-source model race is intensifying and it's genuinely exciting if you care about keeping AI development decentralized.
The 2026 open-source LLM leaderboard is a fascinating snapshot. The top tier now includes GLM-5, Qwen 3.5, DeepSeek V3.2, and Mistral Large, with parameter counts ranging from twenty-seven billion up to a trillion. Chinese labs are competing aggressively at the frontier. GPT-oss 120B leads on general knowledge benchmarks while Kimi K2.5 dominates reasoning and coding. The diversity of competitive models is healthy.
Meta's Llama remains the most widely fine-tuned family of models, alongside Google's Gemma and Mistral's offerings. But there's a problem. A widely shared post noted that most teams fine-tuning these models are doing it wrong. Common mistakes include inadequate data preparation, overfitting to small datasets, and skipping proper validation. The tools are powerful but they're not plug and play.
Microsoft entered with something different. Phi-4-Reasoning-Vision-15B is a compact multimodal model at just fifteen billion parameters that handles math, science, and GUI understanding with images and text together. It's optimized for perception plus reasoning on modest hardware. This matters because not every use case needs a trillion-parameter model. Sometimes you need something that runs efficiently on-premises with strong specialized performance.
Nvidia's CEO Jensen Huang published a rare standalone blog post arguing AI creates jobs rather than destroying them, laying out a five-layer framework for AI infrastructure. And Nvidia reportedly plans an open-source platform for autonomous AI agents, which sent AI-linked crypto tokens rallying and beat the CoinDesk 20 index.
Speaking of AI and legal battles, Anthropic is suing the Trump administration after the Pentagon labeled it a supply chain risk, effectively blacklisting it from government procurement. Anthropic calls the designation unprecedented and unlawful. It's the first U.S. company to receive such a label from the Defense Department. That's a strange situation where one of the leading AI safety companies is being treated as a national security threat.
The broader picture is that open-weight models keep closing the gap with proprietary ones. GPT-5.2 remains dominant in commercial use, but the best open alternatives are now competitive on most benchmarks. That's good for everyone who doesn't want AI development controlled by three companies in San Francisco.
Finally, let's talk about what might be the most consequential technology story that people still treat as science fiction. Tesla is more than doubling its capital expenditure in 2026, spending over twenty billion dollars, and a big chunk of that is going to mass-produce its Optimus humanoid robot.
Musk has been making increasingly bold claims. He says Tesla could be the first company to build artificial general intelligence in humanoid form. The Optimus Gen 3 is entering mass production with a target of up to one million units annually if the technology matures. Morgan Stanley analyst Adam Jonas noted the Gen 3 design might feature surprising simplicity, which would be critical for manufacturing at scale.
But this isn't just a Tesla story anymore. The humanoid robotics field has crossed a threshold. Google DeepMind's Gemini Robotics 1.5 integrates high-level reasoning with real-time motion control. Unitree Robotics has robots interacting with pedestrians on streets in China. The hardware supply chain for high-torque electric actuators is mature and affordable. And the software side has caught up, with perception and behavioral training enabling genuinely autonomous operation in unstructured environments.
What changed is the convergence. Five years ago you had decent hardware or decent AI but not both in the same package at a reasonable cost. Now you do. Tesla's advantage is its manufacturing expertise and its full self-driving AI stack, which translates surprisingly well to humanoid navigation and manipulation.
The financial implications are staggering. If Optimus works at scale, Tesla transitions from a car company to potentially the most valuable company in history. Musk even floated the idea of self-replicating probes in space, which sounds absurd until you remember that the same man runs the company launching most of the world's rockets.
For the Bitcoin ecosystem, this matters because AGI-capable robots would be the most disruptive deflationary force in economic history. If physical labor costs approach zero, the case for hard money as a store of value gets even stronger. Every unit of human productivity that gets augmented by a robot makes scarce assets more valuable relative to everything else.
Here's the thought to sit with. Twenty million bitcoin are now in circulation. Fewer than a million remain to be mined. The ECB is openly worried about losing monetary control to dollar stablecoins. Nation states are both accumulating and liquidating bitcoin based on their fiscal needs. And the technology that will define the next decade, from AI models to humanoid robots, is increasingly open-source and decentralized. The question isn't whether these forces converge. It's whether you're positioned for when they do.