Bitcoin may gain if AI bubble bursts, Hayes says

BitMEX cofounder Arthur Hayes published a new essay, “Situationship,” on Aug. 4, 2026, arguing that the artificial intelligence (AI) infrastructure boom could end as a credit crisis rather than a dot com style equity collapse.

Hayes framed data centers as leveraged real estate containing computing equipment that can lose economic value as newer chips become more efficient.

Hayes said an eventual slowdown in data center construction could expose weak borrowers and financiers, prompting government intervention and broader monetary easing. He believes the resulting liquidity could support a renewed Bitcoin bull market. However, the scenario remains his personal forecast, not a confirmed crisis or an official policy outlook.

Bitcoin traded around $64,150 early on Aug. 5. No evidence reviewed for this report linked the immediate price move to Hayes’s essay. Hayes also acknowledged that he cannot identify the borrower that might trigger a crisis or determine Bitcoin’s precise bottom.

Arthur Hayes says AI spending is a real estate credit trade

Hayes’s central argument is that investors are treating AI capital expenditure as if every dollar supports a high margin technology business. He views much of the spending differently. Data center land, buildings, power connections and cooling systems resemble property development, while processors can become less valuable when newer equipment delivers more computing power at a lower cost.

This distinction leads to his comparison with the global financial crisis. Hayes described the AI boom as a “credit story like 2008 and not an earnings story like 2000.” In his scenario, banks, insurers, private credit funds and infrastructure investors continue financing construction after profitable demand begins slowing.

Losses would then emerge when weaker projects cannot generate enough cash to meet debt, lease or interest obligations. Financial stress could spread to lenders and investors holding AI infrastructure exposure, even if leading technology companies remain profitable.

Hayes expects announced AI capital spending growth to begin slowing during the second half of 2027 and become clearer in 2028. He also expects markets to eventually reward companies that reduce construction plans. Those dates are forecasts. No company filing reviewed for this report confirms that an industrywide contraction has begun.

His Bitcoin case follows from the expected policy response. Hayes argues that U.S. authorities would protect strategically important AI companies and their lenders because computing capacity has become part of the country’s economic competition with China.

He discussed a possible Bitcoin trading range between $60,000 and $70,000, with downside near $50,000, before an eventual rise toward $1 million. Those levels are not guaranteed targets and depend on monetary policy, credit creation and investor demand developing as Hayes expects.

The essay extends an earlier argument. As crypto.news previously reported, Hayes warned that major technology listings, including possible OpenAI, Anthropic and SpaceX offerings, could absorb liquidity that might otherwise enter crypto markets.

In related coverage, crypto.news examined the expanding bond and credit exposure behind AI infrastructure. That analysis noted that financial risks could spread beyond technology shares if data center construction relies more heavily on debt and private financing.

Official filings show AI spending is still accelerating

The latest company results do not show an AI capital spending collapse. Alphabet reported $44.9 billion of capital expenditure during the second quarter. About 60% of its technical infrastructure investment went toward servers, while 40% went toward data centers and networking equipment.

Alphabet raised its 2026 capital spending guidance to between $195 billion and $205 billion, up from its previous range of $180 billion to $190 billion. The company attributed the increase to faster capacity delivery required to meet demand.

Google Cloud revenue rose 82% from the previous year to $24.8 billion. Cloud operating income reached $8.8 billion, while backlog increased to $514 billion. Alphabet said it expects capital expenditure to increase again in 2027.

Microsoft also reported continued expansion. Its quarterly capital expenditure reached $41 billion, with roughly two thirds directed to CPUs and GPUs. Microsoft Cloud revenue increased 27% to $59.3 billion, while commercial remaining performance obligations reached $678 billion.

The company said it expects capital expenditure to grow during fiscal 2027. Microsoft also expects more than $50 billion of spending in its next quarter, although part of that figure reflects a change in how some data center leases will be classified.

Amazon reported a similar mix of rising investment and stronger cloud income. AWS revenue increased 37% to $42.2 billion in the second quarter, its fastest growth in 18 quarters. AWS operating income reached $16.6 billion.

However, Amazon’s trailing twelve month free cash flow moved to an outflow of $7.6 billion. The company attributed the change mainly to a $66.1 billion increase in property and equipment purchases, largely connected to AI investment.

These results cut both ways for Hayes’s thesis. Strong cloud growth and large customer backlogs weaken the argument that demand is already failing. At the same time, lower free cash flow, rising depreciation and growing contractual obligations show how the buildout can pressure finances even while revenue expands.

Heavy spending alone does not create a credit crisis. Such a crisis would require weaker cash generation, refinancing problems, defaults or impaired infrastructure assets across several companies and lenders.

U.S. financing exposure is growing, but 2008 is unproven

Regulatory filings support Hayes’s narrower claim that AI infrastructure increasingly involves leases, guarantees, joint ventures and outside capital.

Alphabet disclosed $85.2 billion of future payments for leases, mainly connected to data centers, that had not started as of June 30. These leases are scheduled to begin between 2026 and 2031, with contract terms reaching as long as 26 years.

Alphabet also reported $811 billion of purchase commitments and other contractual obligations. Most relate to technical infrastructure, inventory, energy agreements and other long term contracts. The company had $98.2 billion of long term debt and issued more than $51 billion of fixed rate notes during the first half of 2026.

Microsoft disclosed $62.9 billion of finance lease liabilities as of March 31. It also reported another $196.6 billion of leases, mainly for data centers, that had not yet commenced.

Meta reported approximately $182.88 billion of uncommenced lease obligations and $237.67 billion of noncancelable contractual commitments as of March 31. The company entered another $24 billion of infrastructure contracts during April.

Private financing is also becoming more visible in U.S. data center projects. Meta and BlackRock announced a venture for a one gigawatt campus in El Paso, Texas. Meta described the project as representing more than $10 billion of investment.

An earlier Meta venture with Blue Owl Capital covered an estimated $27 billion data center campus in Louisiana. Blue Owl funds received an 80% interest, while Meta retained 20%. Part of the outside funding came through debt sold privately to PIMCO and other bond investors.

Meta agreed to lease the Louisiana facilities and provided a capped residual value guarantee under certain conditions. Such arrangements show how data center exposure can be distributed among technology companies, insfrastructure funds, landlords and debt investors.

They do not prove that a 2008 style chain of insolvencies has started. Alphabet, Microsoft, Amazon and Meta remain profitable businesses with large operating cash flows and growing customer commitments. The reviewed filings did not report widespread defaults on AI infrastructure debt or an official government rescue program.

The 2008 comparison therefore remains a stress scenario rather than a present diagnosis. Mortgage losses became systemic because weak lending, securitization, leverage and opaque counterparty exposure spread through major financial institutions.

An AI infrastructure downturn could follow a different route involving unused capacity, falling rental values, obsolete equipment, tenant concentration and long power commitments. Whether those risks become systemic will depend on utilization, refinancing conditions and where losses ultimately settle.

Bitcoin’s outcome depends on policy, liquidity and timing

The Federal Reserve held its federal funds target range at 3.5% to 3.75% on July 29. The decision passed by a 9 to 3 vote. The central bank did not announce an AI rescue facility, emergency lending program or new asset purchase plan.

The Fed has conducted reserve management purchases of Treasury bills to maintain ample banking system reserves. Its July monetary policy report said Treasury bill purchases since early January totaled nearly $250 billion, including about $160 billion of reserve management purchases.

Those operations are not officially described as quantitative easing or an AI bailout. The Fed says they are intended to maintain an adequate level of reserves and support control over short term interest rates.

Hayes interprets balance sheet growth and stable policy rates as supportive for bank credit and future market liquidity. That interpretation remains open to debate because reserve management can expand the Fed’s assets without representing the broad crisis response assumed in his forecast.

Bitcoin could benefit if a future downturn produces rate cuts, emergency lending or larger asset purchases. However, the first stage of a credit shock could hurt Bitcoin as investors sell liquid assets, meet margin calls and reduce leverage.

As crypto.news reported in its examination of Bitcoin’s changing market cycle, Federal Reserve policy and global liquidity now compete with the halving cycle as major drivers of crypto prices.

The next evidence will come from company guidance and credit markets rather than from Hayes’s essay. Investors can watch 2027 spending plans, cloud backlog conversion, data center occupancy, lease commitments, private credit spreads and any defaults tied to AI infrastructure.

The Fed’s next scheduled meeting will take place on Sept. 15 and Sept. 16. Unless company demand weakens or financing stress begins appearing, Hayes’s argument remains a forward looking Bitcoin thesis built around a credit crisis that has not occurred.

FAQs