Freely sharing algorithm indicators and ideas for buy and sell signals. Not financial advice. Have a risk management strategy. Not all alerts follow-through!
⚠️ Nvidia is increasingly using its own balance sheet to keep the AI boom running.
Nvidia is not only selling the GPUs powering the AI buildout, it is also helping finance the infrastructure and customers buying them, including nearly $50 billion invested in frontier AI labs and a planned $105 billion backstop for the Ohio data-center project tied to OpenAI and SB Energy.
Nvidia is also working with Wall Street firms on up to $500 billion of financing for AI chip purchases, while OpenAI could purchase roughly $350 billion of Nvidia chips for the full Ohio buildout.
The strategy can supercharge Nvidia’s revenue as long as AI demand keeps accelerating, but it also increases Nvidia’s exposure if AI spending starts to slow.
If AI labs and data-center developers cannot fund the buildout themselves, Nvidia may increasingly have to support the customers generating demand for its own chips.
The bigger the commitments become, the more painful a demand slowdown could be, because Nvidia would be exposed not only through weaker GPU sales, but also through the financing and guarantees supporting the AI boom.
What happens to Nvidia’s numbers if the AI spending boom materially slows down?
AI-related CapEx is bigger than every CapEx cycle in history.
All previous CapEx manias have resulted in bubbles which eventually popped
This time will not be different.
🚨 WE ARE NOW ENTERING THE HOTTEST PHASE
Every mid-term election year for 50 years has delivered a drawdown
1974 Ford: -35%
1978 Carter: -15%
1982 Reagan: -17%
1990 Bush: -20%
1994 Clinton: -8%
1998 Clinton: -22%
2002 Bush: -34%
2010 Obama: -17%
2018 Trump: -20%
2022 Biden: -27%
2026 Trump: ???
Ten mid-term years. Ten drawdowns. Not one skipped its turn. Average: roughly -21%.
And 2026 has more than the calendar working against it. A new Fed chair, eight weeks into the job. Across nine decades, every new chair was greeted with an equity drawdown in his first three months. Twelve chairs, twelve drawdowns, average roughly -12%. The market doesn't price a person. It prices a probability distribution. And it probes until the new chair reveals himself.
The last time both cycles overlapped: 2018. Powell takes the chair, Volmageddon hits within days, and after "a long way from neutral" the market pushes the S&P down 20% into Christmas Eve. Then Powell blinked. 2019 delivered over 30%. New chair, mid-term year, autumn washout, capitulation low, melt-up. That's the template.
Even fear has a calendar. The VIX troughs in early summer and peaks in September and October, and in mid-term years the crest runs higher. It sat in the mid-teens in early July, right at the seasonal trough. On Friday it jumped above 18. The market has started paying attention. It has not yet paid the full toll.
Meanwhile the shock absorbers are gone. Retail cash allocations at extreme lows seen only in 1998, 2000, 2018 and 2021. Put/call skew at a record low, nobody is hedging. Record IPO supply draining liquidity from the existing market.
And here's the flip: every single one of those ten mid-term drawdowns was a buying opportunity. Not most. All. Since 1934, the average rally off the mid-term low: roughly 47%. The market took out its prior high four times out of five. Bull markets don't die of drawdowns. They die of exhaustion. The mid-term correction is the maintenance schedule of the four-year cycle.
The playbook is not heroic. Hold your quality. Keep dry powder with a shopping list attached, decided at VIX 18, executed at VIX 28. Buy the fear in tranches.
The full map is out tomorrow, free for everyone. Twelve Fed chairs, ten mid-terms, the VIX season, the liquidity cycle, the 1998 rhyme. And the four tripwires that would prove it all wrong.
Greed is obvious. Fear is the edge.
When the cannons fire, buy.
Get ready for the next yen intervention... The yen has reclaimed 160.
In under a month, over 50% of the combined ~$100B intervention has already been reversed.
If this run rate holds, the Fed would have to print ~$50B a month just to stop the BoJ from selling USTs.
To put that into perspective, the post-peak COVID QE pace for USTs was $80B a month.
We don't own enough hard assets for what's coming.
Ray Dalio just said the quiet part out loud.
"If you devalue the money, it makes everything appear to be going up."
The stock market boom is a lie…
What we're really witnessing is the death of the Dollar, not the growth of the economy.
99% of people have absolutely no idea.
🦔Thomson Reuters just built its own AI model for about $40 million over two years. The final training run cost $450,000. They started with Qwen, an open-source model from China's Alibaba, and trained it on their own legal and news content from Westlaw, Practical Law, and Reuters. The company said the move is about reducing its dependence on Anthropic. Their CTO compared paying for outside AI to being a permanent tenant versus owning the building. Enterprise customers broadly have been cutting spending on OpenAI and Anthropic and switching to cheaper alternatives.
My Take
I covered the DeepSeek pricing collapse a few months ago and said the frontier labs would have a hard time defending premium API pricing once open-source got good enough. Thomson Reuters just did exactly what I expected someone to do. They grabbed a free model, trained it on their own stuff, and now they're pulling back from Anthropic. $40 million, done. Anthropic charges that in API fees from a handful of big customers in a year.
The $190 to $200 billion revenue projection Anthropic is selling to IPO investors assumes companies like Thomson Reuters keep paying. They just stopped. And Thomson Reuters put their model on Hugging Face for academics to use, which means the playbook is now public. I think the frontier API business has maybe two or three years before most large companies with good data figure out they can do this themselves, and the ones who move first are going to pressure the ones still paying full price to ask why.
Hedgie🤗
$NVDA core business is 90%+ GPUs, with 60% concentrated in just 4 big clients.
Stripping out uncollected booked sales & equity gains leaves real cash net income at ~$3.3B.😳
That puts NVDA’s true cash P/E:
650x
—overvalued by ~1,500% at current PE & making it more over valued than Enron & Cisco at their 2000 peak.
Here is another example of the vendor-financed demand story:
1. Nvidia holds equity in CoreWeave
2. THEN sells CoreWeave the chips
3. AND backstops them with $6.3 billion valuation
4. WHICH CoreWeave takes to a bank (signed customer contract)
5. TO BORROW AGAINST and buy more GPUs.
Yes, it’s legal but that doesn’t make it any less round tripping!!
AND when $CRWV data center capacity isn't fully used by its own customers, $NVDA is OBLIGATED to buy the unsold capacity through April 2032.
👉So Nvidia sells and funds and backstops the buyer.👈
That’s another way of saying:
DEMAND IS NOT REAL but it is “guaranteed”.