Bitcoin punched back above $65,500 over the weekend after Trump announced a U.S.-Iran peace deal that reopens the Strait of Hormuz, sending oil sliding and risk assets higher. Michael Saylor's Strategy bought another 1,587 bitcoin for $100 million, ETF inflows finally flipped positive on Friday with BlackRock's IBIT pulling in nearly $58 million on its own, and Standard Chartered is calling crypto winter officially over. Meanwhile, Google dropped Gemini 3.5 Live Translate handling 70 languages in real time, Crusoe locked in 5 gigawatts of AI data center contracts, and SpaceX went public, making Elon Musk the first trillionaire and leaving his net worth larger than the entire altcoin market combined.
Let's start with the move everyone's been waiting on. Bitcoin reclaimed $65,500 over the weekend, a two-week high, after Trump said the U.S. and Iran have a deal for what he called a toll-free opening of the Strait of Hormuz. Oil tumbled, equities ripped higher, and bitcoin finally caught a bid after weeks of grinding defensive action.
The context matters here. A ceasefire in April collapsed. A second truce broke on June 9 after U.S. strikes. Both times bitcoin gave back the entire geopolitical move. So traders are right to be a little skeptical of headline number three. Crypto stayed wary even as equities celebrated.
But underneath the price action, the signals are shifting. Standard Chartered's Geoff Kendrick told clients winter is over, pointing to three things: returning spot ETF inflows, falling oil, and Coinbase CEO Brian Armstrong saying his gut tells him bitcoin bottomed near $60,000. Armstrong leaned on the four-year cycle, which has historically marked the lows.
Then there's Saylor. Strategy bought another 1,587 BTC for $100 million at an average price of $63,024. That brings total holdings to 846,842 bitcoin. He raised $209 million through MSTR stock sales to fund it. Whether you love or hate the playbook, Saylor keeps stacking on weakness, and the market noticed.
The counter-case still exists. One historical pattern stretching back to bitcoin's earliest days suggests a flush to $48,000 is possible if certain conditions trigger. And the Fed meets this week, with Warsh chairing his first session. A hawkish surprise could absolutely kill this rally. But for now, the path of least resistance is the $69,000 target traders are eyeing if the peace deal actually gets signed.
Speaking of ETF flows, something worth paying attention to is happening underneath the headline numbers. The U.S. spot bitcoin ETF market is quietly turning into a two-firm race between BlackRock and Fidelity.
Look at the data. On Friday, June 12, spot bitcoin ETFs took in $85.9 million in net inflows, ending a brutal stretch where over $1.67 billion had bled out the prior week. BlackRock's IBIT alone pulled in $57.7 million, about two-thirds of the entire day's inflow. Fidelity's FBTC added $18 million. Everyone else, Bitwise, Ark, VanEck, fought over single-digit millions.
This isn't a one-off. Pull up the big inflow days from this year. January 14: IBIT and FBTC took over 90% of the $840 million daily total. April 17: roughly two-thirds. May 1: nearly $500 million combined out of $629 million. The pattern is consistent, and it's intensifying.
The reason is structural. Advisers, hedge funds, family offices, pension desks, they default to the two biggest, most liquid products from the two biggest asset managers. Tighter spreads, deeper books, brand trust. Once you're the default, you stay the default.
What does this mean for the smaller issuers? Honestly, it's getting grim. The Bitwise, Ark, VanEck, Franklin Templeton products are functional, but they're not moving the market. Capital formation for them is a slog, and you have to ask how many of those funds survive the next bear stretch.
For bitcoin holders, the takeaway is mixed. Concentration in two trusted custodians is fine for institutional adoption optics. But it does mean that if BlackRock or Fidelity ever had to halt creations or redemptions, the spot ETF channel would functionally be broken. Worth keeping in the back of your mind.
Now to mining, where the story this month is about where the hashrate is going and why. Two announcements caught my eye, and they point in completely different directions.
First, Paraguay. HIVE Digital just completed phase two of its Yguazú facility in Valenzuela, hitting 200 megawatts and producing over 8 BTC per day. Their global hashrate is now above 18 exahash. The site is powered by surplus hydroelectricity from the Itaipú Dam, which has 14 gigawatts of capacity. Paraguay uses only about 20% of its share. The rest is essentially stranded clean energy waiting for a buyer, and bitcoin miners are it. Paraguay now accounts for roughly 4.3% of global hashrate, which is wild for a country of 7 million people. HIVE picked up the site from Bitfarms for $56 million in early 2025. Looks like a steal.
Meanwhile in West Texas, Sangha Renewables energized a 20 megawatt mine tied to a 150 megawatt solar farm in Ector County. Behind-the-meter solar, with TotalEnergies providing balancing power when the sun's not cooperating. The pitch is monetizing renewable assets that face grid congestion or volatile pricing, by using flexible mining load as a buyer of last resort. It's a replicable template, and it's the kind of thing that will keep U.S. hashrate competitive even at these depressed hashprice levels.
The broader mining economics tell the rest of the story. Bitcoin difficulty just dropped 10%, the 11th-largest downward adjustment in network history. Translation: weaker miners are capitulating. Hashprice has been at record lows. The miners that survive this stretch are the ones with the cheapest power and the most operational flexibility.
And here's the contrast. Crusoe, which used to be one of the bigger bitcoin miners, has now contracted nearly 5 gigawatts of AI data center capacity. They sold off their entire bitcoin mining operation to NYDIG in March 2025. Their flagship Stargate campus in Abilene is being built with OpenAI and Oracle. They paused a 1.8 gigawatt Wyoming project at a customer's request, but their pipeline is over 20 gigawatts. The bitcoin miners with energy expertise are pivoting hard into AI compute, and that's a competitive dynamic the whole industry is going to have to deal with for the next few years.
On the AI side, Google had a busy week, and one release in particular is going to ripple through a lot of products.
Gemini 3.5 Live Translate launched with real-time speech-to-speech translation across 70-plus languages. The model streams the translated audio just a few seconds behind the speaker, with no need to wait for pauses or take turns. It auto-detects the language, handles noisy environments, and even preserves the speaker's tone and pacing when you use headphones.
The rollout is broad. Developers get it via the Gemini Live API and Google AI Studio. Google Meet enterprises get a private preview this month, jumping from 5 supported languages to 70-plus, with over 2,000 language combinations possible in a single meeting. Consumers get it through the Google Translate app on Android and iOS. Android even has a listening mode that pipes translated audio through the phone's earpiece, no headphones needed, so you can use it discreetly on a tour or in a meeting.
This is one of those features that sounds incremental and then quietly eats an entire industry. Conference interpretation, language schools, international call centers, the long tail of multilingual customer support, all of it gets cheaper and more accessible.
Google also dropped two other things worth flagging. DiffusionGemma is an experimental 26 billion parameter mixture-of-experts model that uses text diffusion instead of token-by-token generation. It can produce 256 tokens per forward pass in parallel, runs in about 18 GB of VRAM when quantized, and generates up to 4 times faster than autoregressive models on GPUs. Quality is lower than standard Gemma 4, so this is for speed-critical local workflows, inline editing, rapid iteration. Not a replacement for the flagship model, but an interesting architectural bet.
And Gemini is now available to Apple developers via the Foundation Models framework. You can swap between on-device Apple models and cloud Gemini with a small code change. Firebase AI Logic handles the backend. Xcode gets agentic, multi-step coding assistance built in. This is Apple finally opening Foundation Models to third-party cloud providers, and Google moved fast to be first in line.
The bigger picture: AI agents are moving from generic chatbots into vertical, workflow-aware systems. Vonage launched industry-specific agents for healthcare, finance, and retail. Decagon's Duet Autopilot claims 93% completion on complex CX tasks. Adidas is running the Audi F1 ecommerce site with a handful of humans and a stack of AI agents. The deployment phase is here. The interesting question is which of these actually deliver measurable outcomes versus which are slideware.
One thing to watch this week: the Fed meets, Warsh runs his first session as chair, and bitcoin is pricing in both a peace deal and a dovish pivot at the same time. Markets don't usually get to keep both of those gifts. Pick which one you think is real.