Welcome to Fulgur News. Let's jump in with a quick look at what's moving. Bitcoin is holding above seventy-one thousand dollars despite geopolitical tensions and stubborn inflation data, with spot Bitcoin ETFs logging their first five-day inflow streak of 2026, pulling in about 767 million dollars. The White House just got Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI to sign a pledge protecting household electricity ratepayers from the cost of powering AI data centers. The EU Parliament reached political agreement on AI Act amendments including a ban on non-consensual deepfakes, with a committee vote set for March 18th. And Blockstream's Jade hardware wallet now supports Lightning Network payments directly from cold storage. A lot to unpack. Let's get into it.
Let's start with Bitcoin. The price is hovering around seventy-one to seventy-four thousand dollars and it's been a surprisingly resilient week. The dollar index broke above one hundred for the first time since late November, oil is flirting with a hundred dollars a barrel after Trump warned of strikes on Iran's Kharg Island, and Q4 GDP got revised down hard to just 0.7 percent from an initial 1.4 percent. Meanwhile core PCE came in at 3.1 percent year over year. That's a stagflationary setup. Higher prices, slower growth. Not exactly a recipe for risk assets to thrive.
And yet, Bitcoin is up over four percent on the week. Strategy, formerly MicroStrategy, added over eleven thousand BTC to its holdings. Spot Bitcoin ETFs had five consecutive days of inflows totaling 767 million dollars, the first such streak this year. That's real money coming back in.
The more interesting macro story though is the gold-to-Bitcoin rotation. JPMorgan flagged a sharp divergence since the Iran conflict began. Gold ETFs have seen outflows of about 2.7 percent of assets, with the GLD fund alone losing three billion in a single day. Meanwhile Bitcoin ETFs brought in around 906 million over thirty days, reversing prior outflows. Bloomberg Intelligence strategist Mike McGlone is even saying gold is no longer a store of value, warning of a 2008-like setup driven by oil shocks and commodity volatility.
Now, not everyone's ready to declare victory. Bitcoin's correlation to tech stocks is still elevated, and the shallow pullback from the seventy-four-five hundred level has some analysts cautioning the bear market may not be over. The Fed meeting on March 17th and 18th is the next major catalyst. If oil above a hundred forces a hawkish shift in rate expectations, risk assets could get hit. But Bitcoin's ability to hold its ground while the dollar, yields, and oil all rise is notable. It's acting less like a leveraged tech bet and more like something else. Maybe not digital gold yet, but certainly not just a risk-on trade either.
One more data point worth mentioning. A Cambridge study spanning eleven years and sixty-eight verified submarine cable failures found Bitcoin's physical infrastructure is far more resilient than previously understood. The network can survive seventy-two percent of the world's submarine cables being cut. TOR adoption has actually strengthened it. The vulnerability isn't the cables. It's hosting concentration. A targeted attack on just five major hosting providers could cripple it. Resilient in the physical world, concentrated in the digital one. Worth thinking about.
Let's shift to AI regulation, because March 2026 is turning into a pivotal month on both sides of the Atlantic.
In the US, President Trump's December 2025 executive order is now reaching its first major deadlines. The order's stated goal is global AI dominance with minimal regulatory burden. The Department of Commerce has been tasked with evaluating state-level AI laws like Colorado's AI Act, California's disclosure requirements, and New York's employment-related AI rules to determine if they conflict with federal policy. The DOJ has stood up an AI Litigation Task Force that could challenge state laws deemed too burdensome, particularly those requiring AI output modifications or disclosures that might raise First Amendment concerns. No lawsuits have been filed yet, but the threat is clearly on the table.
This is a big deal. The US has gone from having no coherent federal AI policy to having one that could potentially preempt state regulations. The direction is clearly pro-industry, aiming for centralized, predictable enforcement through existing agencies like the FTC, HHS, and DOJ rather than creating new regulatory bodies.
Meanwhile in Europe, the AI Act continues its phased rollout and it's getting teeth. The European Parliament just reached political agreement on amendments that include a ban on non-consensual AI-generated deepfake images and eased compliance rules for AI embedded in regulated products like medical devices. These amendments are part of the broader AI Act Omnibus package heading to a committee vote on March 18th.
Here's where it gets really practical for US companies. The EU AI Act applies extraterritorially. Any AI output used by EU customers, employees, or partners falls under its jurisdiction regardless of where the AI was developed or hosted. Violations can trigger fines up to seven percent of global turnover, strict liability under the Product Liability Directive, and potential insurance exclusions. All at the same time. And about forty percent of enterprise AI systems fall into ambiguous high-risk classification categories, making compliance expensive and uncertain.
The transparency disclosure requirements kick in August 2nd, 2026. That's five months away. And compliance isn't a documentation exercise. It requires engineering level changes. Logging architectures, human oversight integration, documentation as code. Many US firms haven't even started. So you have this fascinating divergence. The US pushing for lighter regulation and global competitiveness, Europe tightening the screws with extraterritorial reach. Companies operating in both markets are going to be squeezed between these two approaches for years to come.
Now let's talk about the physical reality underpinning all this AI development. The energy story is becoming impossible to ignore.
The Electric Power Research Institute, EPRI, just released a report projecting that US data centers could consume nine to seventeen percent of the nation's total electricity by 2030. That's more than double their current share. We're talking about going from roughly 177 to 192 terawatt hours in 2024 to as much as 790 terawatt hours by 2030. Individual large-scale AI data centers now consume power equivalent to hundreds of thousands of homes.
The Department of Energy responded this week with a 1.9 billion dollar initiative called SPARK, focused on modernizing and expanding the national power grid. The approach is pragmatic. Instead of building new transmission corridors, which takes years of permitting and construction, SPARK focuses on reconductoring existing lines with advanced materials and deploying technology to increase capacity on what's already built. Project selections are expected by August 2026.
On the private sector side, GridAI and Amp Z announced a partnership to develop and manage over five gigawatts of AI data center capacity across North America. Five gigawatts. That's roughly the output of five nuclear power plants worth of capacity dedicated to AI computing. They're using energy orchestration software to manage grid power, renewables, battery storage, and on-site generation as an integrated system.
But the political question is heating up fast. CNBC ran a deep piece this week on who actually pays for all this electricity. Data center construction hit a record 25.2 billion dollars in January 2026. The concern is straightforward. If utilities need to massively upgrade infrastructure to serve these facilities, do regular ratepayers get stuck with the bill?
The White House moved to preempt that backlash by getting Amazon Web Services, Google, Meta, Microsoft, OpenAI, Oracle, and xAI to sign what they're calling the Ratepayer Protection Pledge. These companies committed to covering costs related to power generation and grid upgrades for their data centers, funding infrastructure, creating jobs, and selling excess power back to grids. Whether this pledge has real enforcement teeth or is mostly political cover remains to be seen. But the fact that the White House felt the need to organize it tells you how serious the energy concern has become.
The fundamental tension is clear. The US wants AI dominance. AI dominance requires massive compute. Massive compute requires massive power. And the grid wasn't built for this. We're essentially trying to bolt a twenty-first century computing infrastructure onto a twentieth century electrical grid while keeping the lights on and the bills affordable for everyone else.
Let's close with some developments in the broader Bitcoin and digital asset ecosystem.
Blockstream quietly made one of the more practical announcements of the week. Their Jade hardware wallet now supports Lightning Network payments. You can make instant Bitcoin payments directly from cold storage without exposing your private keys. This is the kind of unsexy but genuinely useful infrastructure improvement that makes Bitcoin more usable day to day.
On the technical side, Bitcoin Optech highlighted Binohash, a new collision-resistant hash function built entirely with existing Bitcoin Script. Created by Robin Linus, it enables limited transaction introspection and covenant-like functionality without requiring any consensus changes to the network. That's significant because it means developers can build more sophisticated Bitcoin applications using the tools that already exist. No forks needed.
BIP-360 is also worth watching. It introduces Pay-to-Merkle-Root, or P2MR, which is Bitcoin's incremental step toward quantum resistance. Instead of exposing public keys on-chain during spending, P2MR commits only to a Merkle root of scripts. It's cautious, it maintains smart contract flexibility, and it addresses the primary risk quantum computers would pose to elliptic curve cryptography.
In the stablecoin world, momentum keeps building. Circle's USDC overtook Tether's USDT in adjusted year-to-date volume according to Mizuho analysts, who called this significant because the stablecoin winner will be the one people actually use for everyday transactions. USDC's market cap is approaching a record eighty billion, with analysts linking some of the surge to capital flight and turmoil in Dubai's real estate market. Circle also overtook BlackRock in tokenized Treasuries as that market hit a record eleven billion dollars.
Stanley Druckenmiller weighed in saying stablecoins could form the backbone of global payments within ten years, calling them more efficient, faster, and cheaper than fiat on traditional banking rails. And crypto insiders are making the case that while AI developers may not care about crypto broadly, stablecoins are the key to agentic finance, the world of autonomous AI agents making micro-transactions.
The Bitcoin Policy Institute is pushing for a de minimis tax exemption for smaller Bitcoin transactions, targeting August for BTC tax relief legislation. They say bipartisan support is encouraging but warned that time is running out. Practical tax relief would do more for Bitcoin adoption than almost any price rally.
Here's the thing to sit with. The AI regulation divergence between the US and EU isn't just a policy debate. It's creating two different competitive environments that will shape which AI systems get built, where they get deployed, and who benefits. If you're building anything that touches AI, the engineering decisions you make in the next five months will determine your compliance posture for years. And if you're watching Bitcoin hold its ground while traditional macro signals scream caution, pay attention to the structural flows, not just the price. The rotation from gold ETFs to Bitcoin ETFs may be the most important capital reallocation story of 2026. That's it for Fulgur News. I'll talk to you next time.