Hayes Says a $1.5 Trillion AI Squeeze Could Send Bitcoin to $1 Million

Hayes' revised macro thesis: liquidity was created, but AI absorbed it

Hayes' revised thesis is straightforward: money was created, but it went elsewhere. He revisited his framework after Bitcoin fell sharply despite continued money creation, concluding that roughly $1.5 trillion in AI-related debt was issued from 2022 through 2026 and pulled available liquidity into the AI buildout. That helps explain the paradox of Bitcoin underperforming in an environment that should have been supportive. The key point is not that Bitcoin lost appeal on its own; it is that spare liquidity appears to have been redirected.

Bitcoin's base is being tested, not confirmed

Hayes says BTC is building a base in the $60,000 to $70,000 range, with $50,000 as strong support. Bulls can read that as evidence the floor is holding despite weaker liquidity. Bears can argue it is still a pause inside a risk regime where AI has absorbed the flow of capital.

The bullish breakout case only changes if that liquidity balance reverses under stress. Hayes' larger bull case is that Bitcoin could rise above $1 million only if large-scale liquidity injections begin after an AI-led break. That makes the setup more interesting than the current trade.

Why the AI capex boom could become the pressure point

The next question is not whether AI spending is real. It is whether the financing behind it can stay easy long enough for the story to keep working.

The weak link is funding, not demand

Hyperscalers are on track to spend around $800 billion on AI infrastructure in 2026. That is a large absorption of capital, and it shifts the pressure point down the chain. As Hayes has argued, Bitcoin will not truly bottom until the AI bubble implodes and starts to stress the system that financed it. In that view, the market stops rewarding spending on future potential and starts demanding harder proof of economic return.

That scrutiny is already visible in equity prices. Alphabet recently beat expectations for quarterly cloud revenue growth, yet its shares were still down more than 1% in extended trading. The market reaction matters more than the headline beat: strong AI-linked demand alone does not seem to be enough to remove scrutiny from heavy AI spending.

Why slower spending growth could matter fast

UBS sees hyperscalers' capex growth slowing to 25% in 2027 and 6% in 2028. Bank of America's survey also found 82% of investors viewed semiconductors as the market's most crowded trade. Put together, that points to a setup where spending is still large, but the growth rate and market positioning are becoming more fragile.

If that happens, the transmission path likely starts in equities and then moves toward financing conditions:

• hyperscaler spending remains large, but growth in spending slows;
• crowded chip and infrastructure trades lose rerating support;
• investors demand stronger evidence that AI outlays will earn acceptable returns;
• funding stress begins to spread beyond mega-cap tech.

Once that happens, the earlier liquidity thesis gets its catalyst. The BIS has already flagged the sustainability of AI-related investments as a growing financial vulnerability. If the market starts treating AI capex as a burden rather than a free pass, policymakers may have fewer options.

What would actually confirm the Bitcoin setup

For now, the practical move is to watch the market's own receipts instead of chasing the headline.

Confirmation signals

• Recent $273 million in weekly ETF inflows is a better sign than the sustained drains seen earlier this year, but it is still too early to call a reversal. Repeatable spot demand matters more than one strong week.
• Hayes' own behavior matters. Earlier this month he sold Hyperliquid, Near Protocol, Worldcoin and exited Zcash and described the moves as defensive. That points to capital preservation, not conviction that the turn is already here.
• The cleaner confirmation is not more AI enthusiasm. It is stress around AI returns. Cases like Alphabet's strong cloud quarter still getting a negative reaction show that revenue strength alone may not be enough to silence scrutiny.

Invalidation signals

• If ETF rebounds remain isolated and still look small relative to the outflows seen earlier this year, Bitcoin is still effectively liquidity-starved.
• If hyperscaler spending stays firm and the market keeps rewarding heavy AI outlays, the squeeze is not reaching the funding chain in a meaningful way.
• If Bitcoin loses the $60,000 to $70,000 range before any credit stress appears, the floor is failing rather than preparing for a macro bounce.

The decision rule

Hayes says Bitcoin won't bottom until the AI bubble implodes. That makes this a setup to verify, not a trade to assume. Confirmation would be repeatable ETF inflows plus growing market doubt about AI returns. Invalidation would be another weak inflow week, or another quarter in which spending surges and the market still pays up.