Saylor Cites AI-Driven Capital Rotation as Bitcoin Slides 13%; ETFs Log $4.3B in Outflows

Bitcoin’s sharp pullback has revived an old market instinct: find the new thing absorbing risk capital. Michael Saylor’s answer is direct—AI infrastructure is soaking up money that might otherwise support BTC. The rotation lens is the right one to analyze here, because it explains not just price, but behavior across products, treasuries, and narratives.

Start with the tape. Over the last 24 hours, Bitcoin slid about 3.7% to roughly $63,429. It’s down more than 13% on the week after tagging $61,559 late Wednesday, and sits nearly 50% below its October 2025 peak. During the same stretch, the broader crypto market cap dropped over 3.1% to $2.29 trillion, with an estimated $1.74 billion in liquidations; Bitcoin longs accounted for about $635 million.

If there’s a clean barometer for demand, it’s the spot ETF complex. Since May 14, those funds have recorded more than $4.3 billion of net outflows, and they haven’t registered a single positive day since May 13—when they took in around $131 million. Flows are now negative for the year. That pattern looks like rotation, not structural failure: investors are lightening up where liquidity is deepest and reallocating where they see faster cash-on-cash opportunities.

Why AI? Because the buildout is capital-hungry: semis, data centers, power, software stacks. Large allocators often run finite risk budgets. When a new, credibly monetizable theme offers nearer-term revenue than a macro asset like Bitcoin, risk migrates. You can see it in equity proxies—and you can also see the side effects in Bitcoin-linked securities. Shares of the firm (MSTR) have fallen about 15% over the past five sessions to near $128 and are down roughly 30% over the month, while Bitcoin itself is off about 22% in that period. Its preferred stock, STRC—used recently to finance more BTC buys—has slipped below par, changing hands near $95.35 versus a $100 face value. That discount implies a rising cost of capital to keep scaling the strategy.

Skeptics have pointed to another pressure point: the firm’s own sale last week of 32 BTC for about $2.5 million from a Bitcoin position marked at more than $53.8 billion. The absolute size is trivial, but the signaling risk isn’t. In asset classes where conviction is part of the product, even tiny deviations can seed doubt for some holders, especially those who entered via ETFs and think in terms of flow, not keys and cold storage.

There are macro crosswinds too. Geopolitical tension and concerns over higher future energy prices tend to crimp appetite for risk assets. That matters here twice: it can push portfolio VAR tighter and also raises the input cost narrative around both AI data centers and Bitcoin miners, nudging asset allocators to choose the growth story perceived to have clearer immediate cash flow.

So is “AI versus Bitcoin” the right frame? Partly. The ETF data back a capital shift. At the same time, rotation stories can become self-fulfilling when reinforced by daily flows and popular heuristics. Price-insensitive ETF redemptions on down days create smooth supply, while momentum sellers amplify short-term moves. That doesn’t tell you much about long-horizon value, but it does explain the tape.

Where does that leave practitioners? If you believe the Bitcoin investment case is driven by multi-cycle monetary hedging and eventual balance sheet penetration, then episodes like this are duration tests—uncomfortable, but consistent with a market that has added an ETF on-ramp and now lives with the flow-driven reflexes that come with it. If you’re managing around quarterly optics, the rotation remains a credible working model until ETF flows stabilize and energy path uncertainty clears.

Saylor’s message—this is rotation, not impairment—captures the immediate dynamic. The more interesting question is when AI’s capital intensity becomes a tailwind for Bitcoin again: tighter power markets and equity concentration can, over time, revive the portfolio case for non-equity macro assets. Until then, the tape is telling you what allocators are prioritizing right now.