Wednesday, March 25th, 2026. Here's what matters today. Bitcoin is hovering just below $72,000, buoyed by a US ceasefire proposal sent to Iran that knocked oil below $100 a barrel. A massive $14 billion options expiry is coming Friday, and $75,000 is the gravitational price magnet. Meanwhile, BlackRock's Bitcoin ETF empire has blown past $100 billion in assets, the fastest any fund family has ever reached that mark. In healthcare, AI is quietly reshaping drug discovery and diagnostics. And Apple is betting big on on-device AI with new silicon and a Gemini-powered Siri overhaul. Let's get into it.
Healthcare AI has been full of hype for years, but something concrete is finally taking shape. Let me walk through a few developments that paint a bigger picture.
Qualified Health just raised $125 million in Series B funding, led by NEA, bringing their total to $155 million. Founded by Stanford-trained physician Justin Norden, the company isn't building another AI diagnostic tool. Instead, it's solving a very real problem: hospitals are drowning in third-party AI solutions and have no good way to manage them. Qualified Health wants to give hospitals the ability to build and control their own AI, rather than outsourcing everything to vendors. That's a subtle but critical shift. If hospitals can't evaluate, deploy, and monitor AI tools themselves, adoption stalls or, worse, gets reckless.
On the drug discovery side, Insilico Medicine launched PandaClaw, an agentic AI tool that sits inside their PandaOmics platform. What's notable here is the target user: biologists without specialized computational skills. PandaClaw interprets natural language research requests, builds multi-step workflows, accesses large biological datasets, and produces transparent reports. It's essentially giving bench scientists the power of a bioinformatics team through conversation. And separately, Insilico partnered with ASKA Pharmaceutical to apply this AI to gynecological conditions like endometriosis and uterine fibroids, diseases that affect millions of women and have long been under-researched. That partnership is a concrete example of AI addressing unmet medical needs, not just optimizing existing ones.
Then there's Viz.ai teaming up with Alnylam Pharmaceuticals to tackle cardiac amyloidosis, a heart condition that's frequently fatal and massively underdiagnosed because its symptoms look like regular heart failure. Their system uses FDA-cleared echocardiography AI to automatically flag subtle signs during routine echocardiograms, without waiting for a physician to suspect the condition. If you catch amyloidosis early, outcomes improve dramatically. The system triggers confirmatory testing and specialist referral automatically. They're starting with a multi-site pilot and plan to expand to over 2,000 hospitals.
The pattern here is clear. AI in healthcare is moving past proof-of-concept into operational deployment, with tools that address workflow problems, empower non-technical users, and catch diseases humans routinely miss.
Bitcoin is stuck in a fascinating tug-of-war right now. It's trading just above $71,000, having bounced back after the Trump administration sent a 15-point ceasefire proposal to Iran through Pakistan. Brent crude dropped 4.7% on the news, falling below $100 a barrel, and Asian equities rallied nearly 2%. Bitcoin caught the tailwind.
But here's the tension. On the upside, there's a $14 billion options expiry on Deribit this Friday, and the data points to $75,000 as the price magnet, the strike with maximum open interest. Rising open interest across the derivatives market signals growing leveraged positioning. On the downside, 4 separate on-chain metrics show weaker demand. Whale activity is subdued, network growth is declining, and investor distribution patterns suggest caution. Bitcoin keeps getting rejected near $72,000.
The broader macro picture isn't making it easier. US PMI data came in showing growth slowing toward 1% while inflation is creeping back near 4%, reigniting stagflation fears. Rising Treasury yields are pulling liquidity away from risk assets. Meanwhile, gold is posting its longest losing streak since 1920, and the Bitcoin-to-gold ratio has surged 30% since the Iran conflict began. That's a notable shift. Bitcoin is outperforming gold during an active geopolitical crisis, but it's doing so while behaving more like a risk asset than a safe haven. It's repricing geopolitical risk in real time, faster than oil or equity markets react.
The mining side had an interesting event too. A rare 2-block reorganization happened on March 23rd at block height 941,880, where Foundry mined 6 consecutive blocks and AntPool and ViaBTC briefly extended a competing branch. The chain resolved exactly as designed, but it exposed concerns about mining concentration and whether the old 6-confirmations heuristic still carries the safety margin everyone assumes.
Exchange outflows continue to show genuine accumulation. Coins are moving off exchanges into long-term storage. So the picture is: traders are cautious, macro is noisy, but underlying holders keep stacking. Friday's options expiry could be the catalyst that breaks this range one way or another.
Let's talk about the institutional side, because the numbers this quarter have been remarkable.
BlackRock's Bitcoin ETF empire has now surpassed $100 billion in assets under management, making it the fastest fund family in history to hit that milestone. CEO Larry Fink told shareholders that digital assets could become a $500 million revenue generator for the firm within 5 years, putting Bitcoin on equal footing with private markets, insurance, and active ETFs in BlackRock's growth strategy. That framing matters. This isn't a side experiment anymore. It's a core business line.
For Q1 2026 overall, Bitcoin ETFs attracted $18.7 billion in net inflows, pushing total assets beyond $128 billion. BlackRock's IBIT led with $8.4 billion, followed by Fidelity's FBTC at $4.1 billion. Grayscale's GBTC outflows slowed dramatically to just $1.2 billion, a fraction of the hemorrhaging we saw in early 2024. Fee competition has intensified, with expense ratios dropping below 0.25% across most providers. And institutional investors now hold roughly 38% of all spot Bitcoin ETF assets.
But it hasn't been a straight line up. March 24th saw over $200 million in inflows, a nice rebound. But just days earlier, on March 19th, there was a $93 million net outflow, and the 10-day period before that saw nearly $876 million in cumulative outflows. Those were driven primarily by BlackRock and Fidelity redemptions, suggesting even the biggest players sometimes pull back during macro uncertainty.
The 21Shares president, Duncan Moir, made an interesting observation. He sees the next phase of crypto ETFs moving beyond passive index exposure into actively managed strategies. That's significant because it means the product landscape is maturing. We're going from simple Bitcoin exposure to structured products, options-based strategies, and multi-asset crypto portfolios.
STS Digital just launched a structured crypto platform with Kraken as its distribution partner, covering 400 tokens and targeting banks, family offices, and high-net-worth individuals. The infrastructure for institutional crypto is getting built out fast, and it's starting to look a lot like traditional finance's product shelf. Whether that's a feature or a bug depends on your perspective.
Apple is making a very deliberate play in AI, and it's worth paying attention to the strategy, not just the specs.
Let's start with the hardware. The new M5 Pro and M5 Max chips introduce what Apple calls Fusion Architecture, combining two dies into a single system-on-chip. The standout feature is GPU Neural Accelerators that transform the GPU into a dedicated AI engine, delivering over 4x the AI compute of previous generations. Paired with up to 128 gigabytes of unified memory, these chips can run complex AI models entirely on-device. No cloud round-trip needed. For developers, researchers, and creative professionals, the MacBook Pro is becoming a local AI workstation that rivals what you'd get from cloud instances.
On the software side, iOS 26.5 beta reveals the new Siri architecture, and it's a significant overhaul. Apple is using a hybrid inference model. Simple, personal queries stay on-device, processed by Apple's own Foundation Model. Complex, knowledge-heavy requests get routed to Google's Gemini through Apple Private Cloud Compute, with data anonymized before it ever leaves the device. Response times are landing between 850 milliseconds and 1.2 seconds over 5G. There's a new dedicated Siri app that consolidates conversation history, lets you pin chats, search past interactions, and upload attachments for analysis. Developers get access through a new SiriKit GenAI API, enabling cross-app context understanding.
Meanwhile, Qualcomm is working the same on-device angle from the Android side. Their AI Research team has developed a system that compresses reasoning chains by up to 2.4x, shrinking the verbose thought processes of thinking models so they can actually run on smartphones. They used reinforcement learning to cut token usage by up to 8x while maintaining accuracy. The approach uses toggleable LoRA adapters to switch between quick responses and deeper reasoning.
The convergence here is clear. Both Apple and Qualcomm are racing to make AI inference local, private, and fast. The privacy implications are significant, especially for anything involving personal data, health information, or financial details. And it's creating a new competitive axis. It's not just about who has the best cloud model anymore. It's about who can run the best model on the device in your pocket.
Here's the thread connecting today's stories. Whether it's hospitals trying to control their own AI destiny, Bitcoin holders accumulating through geopolitical noise, or Apple pushing inference onto your local hardware, the theme is the same: the value is shifting toward whoever controls the compute closest to the decision. Not the biggest cloud, not the loudest hype. The processing that happens right where you need it. That's the edge worth watching. I'll see you tomorrow.