Bitcoin is hovering near sixty-eight thousand dollars after a whipsaw session where the Supreme Court struck down Trump's tariffs as unconstitutional, only for the president to immediately announce a ten percent global tariff under alternative authority. Markets popped, then dropped, and are now shrugging the whole thing off. Meanwhile, the SEC and CFTC are deepening their joint crypto oversight with a relaunched Project Crypto, the White House is nudging banks toward a stablecoin deal, and a new analysis of eight hundred plus AI agent deployments this year found that seventy-six percent of them failed. Lots to unpack.
Let's start with what's actually happening when companies try to deploy AI in the real world, because the gap between the pitch deck and reality is getting uncomfortable. A researcher calling themselves Neural Minimalist published an analysis of eight hundred forty-seven AI agent deployments in 2026 and found that seventy-six percent failed. Not underperformed — failed. And the reason wasn't the models themselves. It was foundational ignorance. Companies rushing to automate workflows without understanding the technology's actual capabilities and constraints. The article points out that a fifty percent task success rate for an AI agent gets celebrated in trade press as a breakthrough, when honestly it means the thing is a coin flip. That's not production-ready. That's a demo. Now contrast that with where AI is actually working. MinterEllison, a major Australian law firm, built a tool called Lantern that processes legal documents fifty-eight times faster than manual review. Key difference: they scoped a narrow, well-defined problem, invested in data quality, and augmented their lawyers rather than trying to replace them. Review times dropped sixty percent. Similarly, the European Central Bank quietly rolled out AI to transform its Corporate Telephone Survey — the one they use to gauge business sentiment for monetary policy. They're using it for real-time data processing and deeper insight extraction from survey responses. Not flashy, not autonomous, but genuinely useful. Microsoft is pushing utilities to move from AI pilots to production, specifically around unifying IT and operational technology data for grid management. The message from their DTECH 2026 presentation was blunt: stop running experiments, start delivering measurable outcomes. And then there's the other side — Royal Greenland, a seafood company that's been on SAP since 1998, is migrating to cloud ERP and cutting ninety percent of its data volume as a precondition for future AI use. They won't go live until March 2027. That's the unsexy reality. Before you can do anything intelligent with AI, you need clean pipes. So the scorecard right now is: narrow, well-scoped AI tools in specific domains are delivering real value. Broad autonomous agents deployed without deep understanding are mostly failing. The hype cycle hasn't corrected yet, but the data is starting to force the conversation.
Turning to Bitcoin markets — it's been rough. Bitcoin is logging its worst first fifty days to start a year on record. We're looking at back-to-back monthly declines in January and February for the first time ever. The price is clinging to around sixty-seven to sixty-eight thousand. Spot Bitcoin ETFs have shed six point eight billion dollars in outflows, and the average ETF investor is now sitting on a twenty percent paper loss. That's the kind of number where capitulation selling starts to become a real risk. The options market tells the same story. There's a persistent panic premium even as price recovers from the lows. Traders are paying up for crash protection. The put-call structure leans bearish, with some desks pointing to a potential retest of sixty thousand in the near term. Bitwise, on the other hand, says Bitcoin is quote significantly undervalued at these levels. And the extreme fear reading on sentiment gauges would historically suggest a contrarian buying opportunity. One interesting data point from the on-chain side: Bitcoin whales have been quietly accumulating. Since December they've added two hundred thirty-six thousand BTC, essentially offsetting the broader sell-off. Large players are building positions while retail panics. Mining tells a different story of resilience too. Difficulty just jumped fifteen percent, the largest increase since 2021, rebounding to one hundred forty-four point four trillion as hashrate recovered to one zettahash per second. US miners bounced back from winter storm outages, with some offsetting downtime by selling electricity back to the grid. And then there's the quantum computing fear narrative. Charles Edwards at Capriole put out a report arguing Bitcoin should already be discounted for quantum risk and could fall to thirty thousand if the network doesn't show progress on quantum-proof upgrades. Bitcoin developer Matt Carallo pushed back hard, saying Bitcoiners are looking for something to blame for the sluggish price and quantum fears aren't the actual driver. I tend to agree with Carallo. The price action is about macro liquidity, tariff uncertainty, and ETF flow dynamics, not theoretical quantum threats that are still years away from being practical.
On the regulatory front, the ground is shifting fast. The SEC issued guidance clarifying that tokenized securities are still securities, full stop. The format doesn't change the legal status — whether it's on a blockchain or in a traditional database, the same registration and reporting obligations apply. They also opened the door for security tokens to trade directly against Bitcoin without requiring fiat conversion, which is actually a meaningful simplification for exchanges and broker-dealers. Separately, broker-dealers can now classify proprietary stablecoins as readily marketable assets with a two percent haircut for capital calculations. That's a practical green light for stablecoins in traditional finance plumbing. The bigger structural story is the SEC and CFTC relaunching Project Crypto as a joint harmonization initiative. The two agencies are actively coordinating on how to divide oversight of digital assets ahead of anticipated legislation. Ripple's Brad Garlinghouse claims the CLARITY bill — which would define which digital assets are securities versus commodities — has a ninety percent chance of passing by April. That feels optimistic, but the momentum is real. At the White House, there have been multiple meetings between crypto companies, banks, and policy advisors about stablecoin yield. The latest session reportedly made progress, with the administration pushing banks to accept a deal that lets the broader market structure bill advance. The sticking point has been whether stablecoins can offer rewards tied to transaction activity. Meanwhile, Minneapolis Fed President Neel Kashkari went on record calling crypto quote utterly useless compared to AI and dismissing pro-stablecoin arguments as a buzzword salad. Which is a reminder that even as the policy apparatus moves toward accommodation, there are still powerful skeptics inside the central banking establishment. And one more: BTC treasury company executives are formally calling for reform of the twelve hundred fifty percent risk weight assigned to Bitcoin under Basel Three. For context, private equity — the next highest category — carries a four hundred percent weight. That three-to-one penalty is a major barrier to bank balance sheet adoption of Bitcoin, and there's now an organized push to change it.
Let's talk about Bitcoin-native financial infrastructure, because some genuinely novel things happened this week. Ledn completed a hundred eighty-eight million dollar securitization backed by over five thousand four hundred Bitcoin-collateralized consumer loans. This is the first asset-backed securities transaction of its kind. The bonds included an investment-grade tranche priced at three hundred thirty-five basis points over benchmark. The key innovation is automated liquidation of Bitcoin collateral, which gives investors a structural safety net. This is Bitcoin-backed credit entering the same institutional machinery as auto loans and mortgages. Whether you think that's exciting or terrifying probably depends on how you feel about two thousand eight. On the startup side, Voltage launched what they're calling the first payment-volume line of credit with Bitcoin finality and USD settlement. Companies can send payments that clear in seconds over Lightning or on-chain, then repay the credit line in dollars. No forced Bitcoin liquidations, revenue-based underwriting, scales with transaction volume. It's a clever bridge product — businesses get Bitcoin's speed without the balance sheet complexity. The broader venture picture is interesting too. In just five days of February, over four hundred forty million dollars was raised in crypto, but three hundred six million of that was strategic money, not traditional VC. Anchorage Digital took a hundred million, TRM Labs seventy million — both custody and compliance plays. Bitwise acquired Chorus One to build staking and yield capabilities. The money is flowing toward regulated infrastructure, not the next DeFi experiment. Strategy, formerly MicroStrategy, continues its relentless accumulation — they bought another twenty-four hundred eighty-six BTC for a hundred sixty-eight million, bringing their total to over seven hundred seventeen thousand Bitcoin. They're nearing their hundredth purchase. The stock dropped four percent on the news, and their entire stash is currently underwater by about five point seven billion dollars. Metaplanet's CEO Simon Gerovich had to publicly defend the company's Bitcoin strategy and disclosure practices after investor criticism about hidden details on their options trading. The pressure on leveraged Bitcoin treasury companies is mounting as the price stays below cost basis. And one bright spot for Lightning adoption — the Bitcoin Lightning Network exceeded one billion dollars in monthly transaction volume for the first time, with analysis suggesting AI agentic payments could drive the next surge.
Here's the thought to sit with. We're in a moment where Bitcoin's price is having its worst start to a year on record, but the infrastructure being built around it — securitized lending, Lightning credit lines, institutional custody, regulatory frameworks — is the most sophisticated it's ever been. The same pattern is playing out in AI: the hype-driven deployments are failing at a seventy-six percent rate, but the narrow, well-engineered applications are delivering transformational results. In both cases, the signal is the same. The technology works when the foundation is solid. The question isn't whether Bitcoin or AI will matter — it's whether the people building on them are doing the hard, boring work that makes the difference. That's Fulgur News for February twentieth. I'll talk to you tomorrow.