How AI Companies Are Paying for a Trillion-Dollar Habit

Earn CPE with this course

A question has been bugging me. We see these headlines every day. OpenAI raised $40 billion. Meta is spending $65 billion on AI this year. Elon Musk’s xAI pulled in $20 billion in a single round.

These numbers are astronomical. OpenAI’s CEO Sam Altman recently got annoyed when asked how his company, with a reported $13 billion in revenue, plans to pay for its commitments. The company’s CFO even mentioned hoping for a government backstop.

It’s a fair question. How are these companies funding a spending spree of this magnitude without the revenue to match? I decided to dig in. The answer isn’t just about venture capital. It’s a new financial architecture being built on Wall Street.

The Capital-Intensive AI Arms Race

First, it helps to understand what they are spending all this money on. I see it breaking down into three core areas:

• The Science: The foundational research that creates models like GPT-4. This requires hiring the brightest minds from labs like DeepMind and OpenAI.
• The Hardware: The specialized chips and data centers needed to train and run these models. This is where companies like NVIDIA come in, building hyper-specialized GPUs that are incredibly expensive.
• The Energy: Running all those chips takes a massive amount of power. Data centers are causing electricity prices to spike in local communities, and the demand is only growing.

The single biggest lever to pull in all three areas is capital. If you have the most money, you can get the best scientists, the best chips, and the most energy. This is a fundamental shift from previous tech booms. The dot-com era was about software and eyeballs. The social media era used cash flow to build infrastructure over time.

The AI boom is different. It requires billions in upfront investment for physical assets before you can generate significant revenue. This has forced companies to get creative.

Securitizing the Future

So, how do you pay for it? The answer is a classic Wall Street move with a new-tech twist: securitization.

Companies are treating their AI infrastructure like a financial asset. Here’s a simplified breakdown of how it works:

1. Build It: An AI firm spends billions to build a massive data center filled with thousands of high-cost GPUs.
2. Collateralize It: They use that physical hardware as collateral to issue debt, basically taking out a massive loan against their new infrastructure.
3. Package It: They package that debt into something called an Asset-Backed Security (ABS).
4. Sell It: They sell these securities to institutional investors, raising the billions they needed in the first place.

This isn’t a small-time operation. In 2025 alone, $13.3 billion in these AI infrastructure-backed securities were issued. It’s a way to raise huge amounts of non-dilutive capital. Instead of giving away equity, they are borrowing against the future value of their hardware.

Tinfoil Hat Corner: The Data Center Bubble

This new financial machinery creates some serious accounting challenges and potential systemic risks.

The main problem is valuation. How much are these assets actually worth? The GPUs that power this revolution become obsolete incredibly fast. Some companies are using a 5-to-8-year useful life for these chips in their accounting. But in reality, a new, better chip comes out every 1-3 years, making the old ones significantly less valuable.

Michael Burry, of “The Big Short” fame, has been raising alarms about this. If the useful life of these assets is shorter than stated, companies could face massive, unexpected write-downs.

This leads to a bigger, more speculative question: What if we’re building a data center bubble?

We have billions of dollars in securities backed by assets whose value could plummet. If the projected demand for AI doesn’t materialize, or in if a new technology makes current data centers obsolete, we could see a collapse. It feels eerily similar to the 2008 housing crisis, but with data centers and GPUs replacing subprime mortgages as the underlying asset.

The White House has already stated there will be no federal bailouts for AI companies. The risk is squarely on the private sector.

Key Takeaways

After looking into it, a few things are clear:

• A New Era of Capital Intensity: The AI race is being won with capital. The upfront costs for hardware and energy are unlike anything we’ve seen in tech before.
• Financing is Evolving: To meet this demand, companies are creating complex securities. They are essentially borrowing from the future by collateralizing their physical infrastructure.
• Core Accounting is Being Tested: This entire structure rests on the valuation of these hyper-depreciating assets. Miscalculating their useful life creates enormous risk for companies and the investors buying these new securities.

The AI boom isn’t just a technological revolution; it’s a financial one. It’s forcing a reinvention of how massive projects are funded and creating both huge opportunities and potentially systemic risks for the entire economy.

Learn more about EverydayCPE

• Compare Options: See how we stack up against others in our 2025 Flexible CPE Guide
• Understand the Format: Read how Nano-Learning works for CPAs.
• Check Your State: Ensure you are compliant with our State Requirements Guide.

Further Reading

Here are links to recent reports and analysis that cover the specific topics in this article:

• On Massive AI Spending (Meta & xAI):
  • Social Media Today: Meta Commits to Spending $600 Billion on US Projects by 2028
  • PYMNTS: Report: Elon Musk’s xAI Raises $5 Billion and Doubles Valuation
• On OpenAI’s Financial Scale:
  • The New Stack: Making Sense of Sam Altman’s $7 Trillion AI Chips Gambit
  • Entrepreneur: OpenAI Saw More Revenue in Six Months Than All of Last Year (Note: Revenue for H1 2025 was $4.3B, making the gap with spending even larger)
• On Securitization (Asset-Backed Securities):
  • Columbia Threadneedle: Bond investors join the AI infrastructure boom (Note: This report states AI data center securitization hit $17.4 billion year-to-date in 2025)
  • WebProNews: AI’s Debt-Fueled Data Center Frenzy: Risks Mounting in 2025 Boom
• On Michael Burry’s “Bubble” Warning:
  • Business Insider: Michael Burry’s latest argument against hyperscalers… echoes one made by an investor who shorted Enron (Note: This directly mentions his warning on chip depreciation life)
  • The Economic Times: Big Short legend Michael Burry breaks silence… warns another massive bubble is here
• On the “No Bailout” Stance:
  • PYMNTS: No Federal Bailout for AI, Says White House Tech Chief
  • InsideAI: White House advisor Sacks bats down ‘bailout’ talk in wake of OpenAI comments flap