Monday rundown. Bitcoin whipsawed around 81,000 dollars overnight as Iran tensions rattled risk markets, but the bigger story is the money still flowing in. Bitcoin funds pulled in 706 million dollars last week, the sixth straight week of inflows. Square is auto-enabling Bitcoin Lightning payments for roughly 4 million U.S. small businesses starting today. Morgan Stanley is undercutting Coinbase and Schwab with a 50-basis-point crypto trading fee on E-Trade. OpenAI dropped three new realtime voice models with GPT-5-level reasoning. And Strategy quietly bought another 535 Bitcoin days after Saylor hinted he might sell some. Let's get into it.
This is the one that actually matters today. Starting this morning, Square is auto-enabling Bitcoin payments for millions of eligible U.S. small businesses. No setup, no opt-in friction, no merchant onboarding gauntlet. If you take payments through Square, you can now take Bitcoin. The scale here is what changes the conversation. We're talking about roughly 4 million merchants, around 78 percent of Square's U.S. base. Payments run over the Lightning Network, so checkout feels like tapping a card, and Square is waiving all processing fees on Bitcoin transactions through the end of 2026. Merchants can settle instantly to U.S. dollars, which means no volatility exposure, no custody headache, no new accounting line item. Lightspark's David Marcus called it a potential global financial infrastructure moment, and for once that kind of language might be earned. Lightning has been the perpetual almost-there technology of Bitcoin payments for years. Great demos, real volume on Strike and Cash App, but the merchant side has always been the bottleneck. Auto-enabling at the point-of-sale layer skips that bottleneck entirely. Now, the counterpoint. Most merchants will settle to dollars, which means Bitcoin is functioning as a payment rail, not as money on the receiving end. That's fine. That's how Visa works too. The interesting question is what happens when a meaningful slice of those merchants start leaving some balance in BTC because the rail is faster and cheaper than card networks charging 2 to 3 percent. Worth noting, GoMining also announced GoBTC at Consensus Miami, a competing payments protocol settling on Bitcoin's base layer within 12 hours at a flat 0.2 percent merchant fee. Different architecture, same target, the card networks. The payments layer of Bitcoin is finally getting crowded in a good way.
Morgan Stanley is making a serious play for retail crypto, and the pricing is the story. E-Trade just rolled out crypto trading at 50 basis points per transaction. That's 0.50 percent. Compare that to Robinhood at 95 bps, Schwab at 75 bps, and Coinbase's retail flow around 60. Morgan Stanley is the cheapest mainstream option in the U.S. right now. The rollout is in pilot, but it's going to all 8.6 million E-Trade clients by year-end. And this sits on top of MSBT, Morgan Stanley's own spot Bitcoin ETF, which launched April 8 at a 0.14 percent management fee. MSBT just finished its first month of trading with zero days of net outflows. Not one. That's striking. Most new ETFs leak some flow in the first weeks as positioning shakes out. Morgan Stanley's brand and distribution apparently override that. What's going on here is the bank-ification of crypto access. The pitch to a Morgan Stanley client is simple. You don't need a Coinbase account. You don't need to wire funds to an exchange. You don't need to learn what a maker-taker fee is. Log into the same dashboard where you hold your equities and click buy. From a Bitcoin maximalist lens, this is a mixed bag. On one hand, it's exactly the institutional plumbing that drives the next leg of adoption. On the other hand, it's custodial, it's wrapped, and it pulls demand away from self-custody. The good news is the people Morgan Stanley is converting were never going to run a node anyway. They were going to buy gold or sit in cash. Now they're buying Bitcoin through an interface they trust. Coinbase still has the active trader market and the on-chain functionality. But for the passive retail allocator, the banks just became more competitive than the crypto natives. That's a real shift.
OpenAI shipped three new realtime voice models, and this is the clearest signal yet that voice is becoming a first-class interface for AI agents. The headline model is GPT-Realtime-2. OpenAI claims it brings GPT-5-level reasoning to live conversations. It can reason while talking, call multiple tools in parallel, handle interruptions gracefully, and the context window jumped from 32k to 128k tokens. The Big Bench Audio benchmark went from 81.4 percent to 96.6 percent. That's a step change, not an increment. The other two are GPT-Realtime-Translate, which does live translation across 70 input languages into 13 output languages at 3.4 cents a minute, and GPT-Realtime-Whisper for live transcription at 1.7 cents a minute. Early testers include Zillow, Priceline, and Deutsche Telekom. The use case that matters here is what OpenAI calls voice-to-action. You speak a task, the model reasons through it, calls the tools it needs, executes, and reports back, while audibly telling you it's working so there's no dead air. That's the missing piece for voice agents. The previous generation went silent during tool calls, which broke the illusion immediately. Worth pairing this with ElevenLabs' Expressive Mode 2026, which sits on top of their TTS layer and governs emotional tone and turn-taking. Frustrated user gets empathy, delighted user gets energy, uncertain user gets clarity, without explicit prompting. Combined response latencies are under 250 milliseconds end to end. The competitive picture is sharpening. OpenAI is going wide and developer-first. ElevenLabs is going deep on voice quality and conversational dynamics. Google Gemini Live is in the mix. The companies building actual voice agents now have three serious vendors to choose from, and the quality bar is high enough that the demos finally look like the product. The applied AI rollout in customer support, healthcare intake, and sales qualification is going to accelerate hard over the next six months.
Strategy is back to buying. Michael Saylor confirmed last week that the company purchased 535 Bitcoin for 43 million dollars, funded by common stock sales. This comes days after Saylor floated on the Q1 earnings call that Strategy might sell some Bitcoin to fund dividend payments or offset taxes. The signal whipsawed sentiment. Some holders read it as Saylor capitulating. Others read it as tax loss harvesting, the same playbook Strategy ran in 2022. Saylor's latest tweet basically confirms the latter. The mechanism is straightforward. You sell tranches purchased at higher cost basis to realize losses, use those losses to offset gains elsewhere, then immediately re-accumulate. You end up with the same Bitcoin exposure and a lower tax bill. It's not weakness, it's accounting. Meanwhile, Capital B raised 17.8 million dollars, including from Adam Back and TOBAM, to add up to 182 Bitcoin to its treasury. Smaller scale, same model. The treasury company playbook is now well-defined enough that mid-cap versions are getting funded routinely. On the institutional side, Canton Network's parent Digital Asset is raising 300 million at a 2 billion dollar valuation, led by a16z crypto. Circle pulled in 222 million for its Arc blockchain token at a 3 billion valuation, backed by BlackRock, Apollo, and Bullish. And Ripple raised 200 million from Neuberger Berman to expand Ripple Prime. These are not retail rounds. These are institutional capital pools writing nine-figure checks into crypto infrastructure plays. The pattern across all of it. Tradfi is no longer dipping a toe. Morgan Stanley is launching products. BlackRock is anchoring token sales. Neuberger Berman is funding crypto prime brokerages. The infrastructure layer is consolidating around credentialed counterparties, and the capital is flowing in faster than most people watching the price chart realize.
One prediction. By the time we hit Q4, the question won't be whether banks offer Bitcoin. It'll be why anyone uses a crypto-native broker for spot exposure when their existing bank is cheaper, simpler, and already has their money.