The AI trade just got audited: a 12.6% Alphabet drop, a 28.58% Intel slide, and why lenders and insurers are now part of the story

Free cash flow, financing capacity and whether data-center spending pays back were.

From equity story to credit story

Three data points show the question leaving the stock market:

• OpenAI reportedly cut its fundraising target from $30 billion to about $20 billion, and its target valuation from $1.4 trillion to $1.25 trillion.
• Barclays is exploring credit protection on more than $10 billion of AI infrastructure loans.
• Some insurers are approaching their exposure limits for data-center risk.

The last point is easy to skip and may matter most. Data centers are long, debt-heavy builds. If insurance capacity tightens, protecting those loans costs more, and financing new projects gets harder even while demand for compute stays strong.

Why the supply chain is the weak link

The exposure does not stop at the developers. Contractors, equipment makers and materials suppliers borrow too, often spending on labor, inventory and capacity before they are paid in full. If a project is delayed, cancelled or earns less than planned, payments slip down the chain: contractors wait, equipment orders get pushed out, suppliers sit on stock.

Banks lending to those suppliers can end up carrying AI risk without ever lending to an AI company directly. That is how a valuation question turns into an ordinary commercial-lending question.

What this week tests

Bitunix points to three events: more tech earnings, the US midterm elections, and Friday's nonfarm payrolls. If companies keep raising capex without showing stronger free cash flow, the market may stop paying a premium for spending alone. If jobs data holds up, demand looks resilient, and a rotation inside the AI trade need not mean the cycle is over; money may simply favor firms with better cash generation and tighter capital allocation. Weak payrolls would sharpen worries about financing conditions.

Where crypto sits

Crypto does not get a pass. If AI-financing worries turn into a broad cut in risk exposure, digital assets could see more volatility. If earnings, jobs and demand stay firm, money can rotate within risk assets without a general sell-off.

Here is how the relevant Bitunix perps looked at 10:27 AM ET today. Bitunix quotes funding per interval, and every pair below settles every 8 hours:

• $GOOGL.USDT: 286.69, up 1.11% from its 24h open of 283.55. Funding 0.0254% per 8h.
• $INTC.USDT: 64.64, down 1.88% from 65.88. Funding 0.0340% per 8h, the highest of the group.
• $NVDA.USDT: 210.10, up 2.35% from 205.28. Funding 0.0006% per 8h, close to flat.
• $BTC.USDT: mark 77,950.1, funding 0.01% per 8h.
• $ETH.USDT: 2,291.54, up 1.04% from 2,268.04, funding 0.0043% per 8h.

In cash terms, a 1,000 USDT long on the Intel perp pays about 0.34 USDT per 8-hour settlement while that rate holds, roughly 1 USDT a day. Positive funding after a 28% collapse means longs are still paying to hold a rebound, a position that can unwind fast if the next guidance disappoints. All of these numbers move by the minute.

A short checklist for the week

1. Before each print, write down what would change your view. Higher capex guidance without better free cash flow is the version the market just punished.
2. Stock perps react around US earnings hours while BTC trades through the night. A gap at the next cash open can skip past a stop.
3. Size so that one more double-digit earnings move does not take most of your margin.
4. Treat Friday's payrolls as a second event, not a footnote. The credit side of this story depends on financing conditions.

The shift is simple to say and hard to price: the AI story has moved from how fast companies can build to who is carrying the debt if it does not pay.