Title: The Worm Has Turned — NVDA, AI, ORCL, Positioning Update, and Uranium Show: Paulo Macro (Substack) — paid Guest: Paulo Macro ("Cloudbear") Date: 2025-NOV-24 URL: https://paulomacro.substack.com/p/the-worm-has-turned Length: written post (no timestamps) Note: Back-filled post (processed 2026-JUL-07; predates most of the source's archived posts). The AI/NVDA/OpenAI/datacenter narrative has decisively turned (Gemini 3 leapfrog, NVDA receivables anomaly, ORCL CDS blowout); a positioning update (still room for downside); and a bullish uranium aside (buying again, SUR/floor-price catalyst). Body reproduced for personal study; Substack chrome removed. The appended "The Information" article (Nov 20) is retained as it was included at the foot of the note.
Hope everyone managed to enjoy a nice weekend after such a wild week.
We spent a lot of time reviewing market positioning, sentiment, and valuation last week in Wednesday's Better Not Disappoint. Considering the extreme nature of those readings, this market will ultimately require either a lot of time or a lot of downside to properly rinse. Still, with so much circulating around the AI Narrative and initial shakeup in risk, it's hard not to feel a little lost especially given it's now November 24th… do we bounce? Do we go for real downside with little relief in order to force a Fed accommodation like 4Q 2018? Will the Fed response be a news failure anyway like rate cuts in late 2007? Do the wiggles even matter considering the extremes we reached in October in positioning, valuation, and market concentration? Is it Risk Off across the board from here, or a fierce rotation similar to what we saw on Friday where NVDA becomes the ultimate funding short for Boomer Value longs as my friend Le Shrub now calls his portfolio? Remember in July 2024 a few weeks before Yenmaggeddon when I walked through scenarios for what things might look like if correlations rose and Dispersion Bros lost control in What the Dispersion Trade Unwind Could Look Like? Will we do a 2.0 and see a +5% Russell, -5% QQQ day during these upcoming holidays amidst an illiquid tape, just so P&L Search & Destroy shakes the tree as PMs try to save their year? After all, the kids want the GI Joe with the kung fu grip for Christmas.
Well, I'm openminded…
Quick Housekeeping on Uranium
For those subscribers involved in uranium, on Friday morning I joined my pals Justin & Richard at Uranium Insider's monthly webinar for an in-depth discussion on how I'm seeing the uranium market. They have graciously provided a link to my subscribers (link here, password is UIPaulo), and I would recommend you take a look at their service.
This uranium correction has been frustrating, but not altogether surprising as I warned that "even pretty girls get hurt in a bus crash." We know uranium has been AI and power adjacent while sitting in the quant momentum factor baskets, particularly with OKLO meme-type garbage driving the key URA uranium ETF, so naturally there would be crossfire to miners. The good news is the spot price seems to have stabilized despite the late-week Risk Off in public equities as Cameco and at least one utility showed up to buy spot at $76-77/lb, which they really should be buying given the term price is sitting ~$85 and the carry trade is wide open. We know Cameco has been borrow pounds to meet deliveries — a game that can be played for only so long. The spot price is in a terrific place for them to buy to deliver into contracts and replenish inventories. Unless the world ends and lines to traders are pulled everywhere, I think there is a window here ahead of the next government shutdown on ~January 30th where the government will jam through more critical minerals deals, and I feel pretty highly confident a Strategic Uranium Reserve (SUR) will be announced in the coming months, along with possibly a triple digit floor price for US-based production along the lines of what we saw with Mountain Pass rare earths. Uranium equities need a "wedge" to firmly separate them from the AI/data-center/power world, and this could be that wedge. There has been too much smoke around this for there not to be a fire.
I saw values reemerging on some of the equities as well as event/catalyst risks, and I was buying again on Friday.
The Worm Has Turned in NVDA & AI
The narrative around NVDA, OpenAI, and data centers is in serious trouble. I appreciate the temptation to conclude that all the negative press over the past 72 hours around NVDA's share price and the Google Gemini GPU terminator might suggest contra trades, but I think people are losing sight of the forest for the trees. Just because bubble decriers have spent two years being run over by a bubble doesn't mean there won't be blood when the turn comes, and in narrative space, the turn is real. Once you are on the backside of the narrative and flow mountain, it is a long way down in time and price. Trade the wiggles if you must, but stay focused on what matters. Still, we all know there will be furious short-covering relief rallies along the way, and the challenge as mentioned at the top is what if we get one as a function of last week's selling?
To set the stage, I had a long summary half-written on Saturday, and then realized I had not read my BFF Le Shrub's latest note from Friday morning called "AIBS Part VI — KISS" and then realized damnit, he basically wrote everything in a better way than I ever could and beat me by two days to boot :-) ... To save me some time and effort, Shrub has graciously provided my subscribers with a private link to his note. Please read it and consider subscribing to him — I have no skin in his game other than giving thanks every day for his personal friendship, life advice, and trading counsel (it is the season of giving thanks), so I recommend him open heartedly. In addition to being hysterically funny, he really is a black belt in The Game.
It will be an injustice to his work, but I will add a few points here to explain why the worm has firmly turned.
First, Google's Gemini 3.0 was the biggest news last week … not NVDA's result, although that was a close second which I will discuss briefly below. Gary Marcus published a quick summary worth reading on Wednesday several hours before NVDA's quarter came out. It was his note that got me to short the NVDA pop after hours (mentioned on the substack chat). The bottom line is Google has caught up and passed Open AI, and they didn't need NVDA's GPUs or a ridiculous Stargate/Oracle/Blue Owl datacenter financing morass to do it. Sam Altman was already aware of this, which is why he wildly overreacted to a billionaire investor in his own company on a podcast the week before. Shrub and I discussed this on our Monday Fly on the Wall — it wasn't just hubris, it was the natural reaction of powerful rich creatures in their habitat when they are unexpectedly under threat… just like a tell at the poker table… Sam Altman knew that Gemini 3.0 had him in a corner a week before the results came out.
Then on Thursday, The Information dropped a bomb about Altman's internal memo to OpenAI employees warning of "rough vibes" and tough sledding ahead. The article is included at the end of this note in case you missed it. The key difference of course is OpenAI has no culture compared to Google, which for all its twists and turns has institutional memory and internal flexibility which wins over single-minded obsession and key man risk. That's why its chart looks like this…
… and that's also why I debate Risk Off vs Rotation above. Is it all a GOOG/NVDA pair trade, or is Jevon's Paradox about to bankrupt most players now that they all raced to mobilize billions in credit to do something that will be cheaply accessible to all?
The narrative around the NVDA/OpenAI/datacenter nexus only got worse all week. Masaponzi sold his Nvidia stake, and because he is the close cousin to giant SWF money, instead of buying Google he bought… OpenAI! Yet Berkshire Hathaway bought Google in Q3 as it was screaming to new all-time highs… these are not the same.
The FT has been on fire in the AI realm recently — their work on OpenAI's inference costs running far ahead of implied revenues suggested two weeks ago that the math would simply not line up.
A Missionary Statement from Nvidia's High Priest
A word on the NVDA result… like month ends, I start drinking early on NVDA reporting dates because really, what else is there to do. Forgetting just how toxic Twitter can be, I was amazed at the vitriol that came back for flagging the oddity between YoY and QoQ revenue growth vs accounts receivable growth. I called it "disproportionate," and the tweet wasn't even about NVDA but rather how in the that noise everyone had missed that Blackrock CLOs had failed collateral tests…but boy did the responses pile on. The irritation notwithstanding, I slept like a baby and knew I had more NVDA selling to do at the open because as I have mentioned before, it is maybe only twice a year that a tweet creates such a ruckus that I have to triple a position sight unseen (e.g.: buying Cathie garbage in June 2022, selling bonds in July 2023, cutting down uranium in Dec23, crypto Max Stupid in Nov24, etc).
The receivables anomaly has since been described by others, with one of the better summaries here. My buddy Brandon Beylo of Macro Ops then dropped some knowledge with this tweet that took cash generation back to basics. The TL/DR is that Nvidia is very clearly increasing its cash consumption despite its revenue growth, which is putting a massive strain on its working capital and balance sheet. There is no denying this. It should not be controversial. The fact that it has been vehemently opposed under the rubric of "haters just don't get it" is a very big tell that this stock is going to be in trouble with a lot of downside once the worm turns...
But mafs are not enough. On Friday we saw reports that Jensen himself was aware of all this, and did his own GE Jack Welch impersonation on how to thread the needle between beating the quarter enough to avoid disappointing expectations and bursting the bubble, while not showing so much growth that they make it obvious that their funding of clients to buy chips means it's a giant bubble.
Of course, Trump can be counted on to try to save his primary scorecard that is the stock market, and so news leaked on Friday that White House staff were recommending NVDA be allowed to sell H200 chips to China — yet the stock still closed down in a second back-to-back "news failure" day. The first thing that popped into my mind: The boss move by China would be to say "actually we figured it out on chips — thanks though."
The trader in me requires that I keep a diary of key events in an unfolding narrative. The market price around big events tends to act as an eventual magnet that must be revisited, even if the market has long forgotten. Think of "gap fills" like the recent ORCL move.
Another example: some of you will remember my obsession a year ago (Please Not Another Bitcoin Post) after I flagged that on November 13th, 2024 that the Hawk Tuah meme girl publicly thanked Michael Saylor for getting her into bitcoin at $93k (a post that got her 1.1 million views in a few days, since deleted), while on the same day NFL wide receiver Odell Beckham bragged about how he took his salary from the Rams in bitcoin. That $93k line has perversely acted like an average price magnet since.
Does anyone remember when NVDA's Jensen signed a woman's breast on 4 June 2024 at the Computex conference? Sure it took almost 18 months for the narrative to turn, and I was certainly a "top caller" then… but for context, besides serving as the average price magnet until after Liberation Day as the stock went nowhere for almost a year, I suggest now that at some point this price will need to be revisited, and it had really better hold. Ironic that the Michael Burry family office is long $110 puts…
The Oracle: There is no Spoon
The writing for ORCL has been on the wall for weeks. We have all seen the Oracle CDS chart too many times to count... and while the real damage appears confined to ORCL credit so far this month, the other hyperscalers are not exactly rallying.
Again, there were signs the narrative was starting to flip. The FT ran a damning piece on Oracle on Tuesday, with some eye-opening charts. Deutsche shows just what an outlier Oracle has become. While ORCL may yet end up as the Worldcom posterchild for this cycle's datacenter credit SPV insanity (just don't tell me they went full-blown Star Wars — sorry, Stargate — and called an SPV "Chewco" …google it if you weren't in the markets then), the rest of the sector has not exactly shown restraint.
The debt has moved off balance sheet to the ABS market in classic opaque disintermediation, just as discussed nearly 2 months ago on October 1st with my pals Tony Greer and Jared Dillian (weeks before the hyperscalers went into overdrive on issuance and Jamie Dimon used the term "cockroach" in credit). The numbers will not math. The Information also warned on revenues vs capex on Nov 9th. Nomura also piled on last week. But some of the more measured, balanced minds in Silicon Valley are already taking their AI capex projects lower… much lower (Josh Wolfe).
Positioning — Something for Everyone
So we finally got a whiff of fear last week… what do we do here? Mag6 Put/Call Volume spiked on Thursday's selloff. But zooming out for context over the last five years, there is a lot of room for people to buy puts for this to look remotely like 2022-23, which wasn't even that nasty a bear market (and certainly did not have a massive private credit downcycle layered in). Put/Call 10-day trailing average has risen, but is only just entering the 2015-20 and 2022-23 regime ranges. If we have changed regimes, rather than a mere correction in trend, there is room to go.
Junk credit has been leaking since September after being much tighter in 4Q24-1Q25… what a divergence (and room to widen to Yenmaggeddon or Liberation Day levels). Bloomberg was flagging the cost of carrying margin debt this week as a % of GDP. Deutsche flagged the change in margin debt — this is a problem that will take time. More concerning — and something I flagged in Wednesday's note — is retail continues to show signs they are running out of ammo (Vanda). There are a lot of leveraged ETF inflows to potentially unwind. Yet give it another week and retail historically comes back … but for ETFs. Single stocks? Not so much…
Systematic/CTAs have derisked some, but again — a lot of room to go, even on a 1-year basis. Yet hope springs eternal — Bluekurtic flags that this is the 9th worst first 12 trading days of November on record, yet the 20 weakest starts show the market has often bounced back strong (great, seasonality again). And you can see the hope lingers in Helene Meister's weekly poll this weekend where bulls outnumber bears for 6 straight weeks. The problem is the context of a potentially broader Risk Off.
Correlations are rising and Dispersion Bros are starting to lose control as mentioned at the start, with much more room to go before we hit Yenmaggeddon Aug24 or Liberation Day stress levels (or 2022-23 regime bear market levels), even if 1m-3m correlations are hitting peaks — the absolute levels are still low.
In volatility, we can see that the inversion of 1m over 3m VIX could suggest a near-term low that augurs a relief rally (again, barring an outlier like Yenmaggeddon or Liberation Day). Still, forward vol in the 20s remains constrained because of upcoming holidays (it gets quiet right? But it also gets illiquid…), and this low implied volatility is still below correction levels like Yenmaggeddon or Liberation Day, never mind the persistent implied forward vols in the 30s back in 2022-23. Liquidity remains challenged.
Besides zooming out, I think to get a fuller picture we need to keep some context of other risk assets in mind. For starters, we saw Palantir CEO Alex Karp sold 585,000 PLTR shares for $96 Million on Thursday…don't forget Karp had his own hubristic defiant appearance on CNBC two weeks ago.
But the biggest one is how illiquidity and tech momentum between NVDA and bitcoin are being forged. The damage in crypto is shaping up to be an avalanche of unrelenting bad news — here's a sample: BitMine Immersion Sitting on $4B Loss on Ether Bet as Analyst Warns of Structural issues (Coindesk). The company reported $328 million in profit for fiscal 2025, but it faces structural issues, 10x Research's Markus Thielen warned. Investors could be trapped in an opaque, costly structure as high compensations, lackluster ETH staking yield and disappearing NAV premium linger. BitMine's leadership compensation and external advisors could extract $157 million per year over 10 years through compensation and advisory contracts.
Yes bitcoin is oversold while seeing some dramatic outflows. The problem with the current oversold condition in bitcoin and calling for relief rallies is the terrible location where 71% of invested capital in the entire asset class is underwater — that's how much has traded in as bitcoin went "pro." BofA rightfully asks if BTC has reached "peak global liquidity" now that a record number of central banks have cut rates… keeping in mind that Goldman's FCI is barely off the lows.
Bottom Line
There are reasons to think equities could bounce, but between the NVDA narrative rolling over and the broader context of market liquidity risk, the extreme nature of readings in October, and looking farther back than a year's worth of history, I think one should be very careful. As for the idea that a giant rotation will work — I think we are seeing it in Value and there are pockets where event-driven could withstand a bus crash, but small caps are a problem because the nature of the Russell 2000 has turned into such garbage at the top that I could see a few days of violent outperformance suddenly converge on the downside, particularly if the Fed stands pat in December or does another "hawkish cut."
It's worth remembering that the market is still up 12% on the year and investors have not yet had their feet put to the fire. Think about what things will feel like for managers in two weeks if the market is only +5% YTD, and now you have to ask yourself "do I buy here and get run over, or do I save my year and watch some gnarly Santa Rally take it to +12 again and get me fired for underperforming?" Now that's a painful discussion.
Stay frosty…
As always, kindly yours,
Paulo aka Cloudbear
---
APPENDIX — OpenAI CEO Braces For Possible 'Economic Headwinds' in 'Catching Up' to a Resurgent Google — The Information, Thursday 20 November 2025
OpenAI CEO Sam Altman told colleagues last month that Google's recent progress in artificial intelligence could "create some temporary economic headwinds for our company," though he added that OpenAI would emerge ahead. After OpenAI researchers heard that Google had created a new AI that appears to have leapfrogged OpenAI's in the way it was developed, Altman said in the memo that "we know we have some work to do but we are catching up fast." Still, he cautioned employees that "I expect the vibes out there to be rough for a bit."
The memo foreshadowed Google's launch this week of Gemini 3, an AI model that software developers say excels in automating tasks related to website and product design as well as in coding — a capability that is one of the most important drivers of revenue at AI firms like OpenAI.
Altman's comments show that OpenAI's technological lead over rivals like Google and Anthropic has narrowed. Investors have sunk more than $60 billion into OpenAI, recently valuing it at $500 billion, on the belief it will continue to dominate the market for developing AI that creates content and reasons the way humans do.
That domination is teetering. Anthropic, a four-year-old firm whose founders previously worked at OpenAI, appears poised to generate more revenue than OpenAI this year from selling AI to software developers and businesses through an application programming interface, The Information reported earlier this month.
Google, meanwhile, continues to leverage its search engine to attract more people to use its Gemini chatbot, which competes with OpenAI's ChatGPT. To be sure, ChatGPT is significantly ahead of the Gemini chatbot in terms of usage and revenue, but the gap has been shrinking. "ChatGPT is AI to most people, and I expect that to continue," Altman said in the memo.
Google's other advantage is economic. OpenAI is one of the fastest growing businesses in history, going from next to no revenue in 2022 to a projected $13 billion this year. But it also has projected it would burn more than $100 billion in pursuit of human-level AI in the coming years while spending hundreds of billions of dollars to rent servers to do it, meaning it will likely need to raise the same amount in additional capital. Meanwhile, Google, valued at $3.5 trillion, generated more than $70 billion in free cash flow over the past four quarters alone.
The financial disparity between OpenAI and established firms like Google has prompted public market investors to question whether the startup's unprecedented revenue growth will be enough to erase concerns about its future cash burn. Altman, in his note, acknowledged that "by all accounts, Google has been doing excellent work recently," especially on pretraining. He said he wanted to focus on "very ambitious bets" even if it meant OpenAI would get "temporarily behind in the current regime," and ended: "we are doing remarkably well as a company…and I expect that to continue."