Oracle and OpenAI signed a 5-year mega-scale cloud and AI infrastructure contract worth $300 billion (about 420 trillion won).
Both are currently loss-making companies.
Open AI must pay $60 billion annually to Oracle starting in 2027, but its current annual sales are only $13 billion → 5-fold growth in sales is required within two years.
Oracle must first spend $150 billion to build a data center before receiving that money, but its cash flow is already at minus $10 billion and its credit rating is at risk of being downgraded.
→ If one side is crooked, both fall together.
2. December 11, 2025, the day the market first realized
When Oracle announced that it would increase capital expenditures (capex) by an additional $15 billion after announcing its earnings, the credit default swap (CDS) spread soared to 141bp (the highest level since the 2008 financial crisis).
The stock price plummeted 11% in one day, and the market capitalization of $70 billion evaporated.
Now, Oracle CDS has virtually become “an indicator for hedging the entire risk of AI infrastructure.”
3. ‘Infinite Money Glitch’ where money spins around
1. NVIDIA → Invest hundreds of billions in Open AI
2. Open AI → Use that money to purchase NVIDIA GPU + Oracle Cloud in bulk
3. Oracle → Use that money to buy NVIDIA GPUs and build a data center
4. Nvidia stock price rises → investment in Open AI possible again
→ Same structure as Cisco and Lucent lending money to new telecommunication companies and selling equipment during the telecommunications bubble in 1999-2000.
4. Even Softbank is involved here.
Masayoshi Son bets a total of $113 billion on OpenAI + Stargate (Oracle-OpenAI joint data center project)
However, actual cash is only $58.5 billion → $54.5 billion shortfall
When the value of Open AI falls, Softbank also falls along with it.
5. The scariest thing: “Companies don’t use AI that much.”
MIT, McKinsey, and Gartner survey results → 80-95% of companies responded, “I have spent money on AI so far, but it has had no impact on profits.”
The crazy growth prospects of Open AI and Oracle are based on the premise that “companies will suddenly pour trillions of dollars into AI,” but in reality, that is not happening at all.
conclusion
“AI may be a real innovation, but it may take 10 to 15 years for that innovation to be commercialized, and investors have poured $2 trillion into it, expecting returns within three years.
This structure of round snakes (Oracle-OpenAI-Nvidia-Softbank) biting each other’s tails will collapse if even one side falls.
There is no government relief. “The market will judge.”
To put it simply:
“Spending money on AI infrastructure now could be more dangerous than the 2000 Internet bubble and 2008 financial crisis combined.”
https://substack.com/inbox/post/181416869