- The Opening Print
- Posts
- Every Dog Has Its Day
Every Dog Has Its Day

THE OPENING PRINT
Our view.
// OUR VIEW | — |
The ES and NQ did rally, but NVDA fell 4%, dragging the markets down again. The ES opened 45.25 points lower, and the NQ opened 365.75 points lower.
As I have always said, the public is always sold a bag of goods. This time last year, CNBC, Bloomberg, Fox Business, and hundreds of other news outlets were all about the tech/AI boom, all the billions these companies had made on earnings, and all the billions they were going to spend. But fast-forward to today, and the new reality looks more like the world's largest Ponzi scheme. I came up with this idea myself. I didn't read it anywhere, I didn't take it from anyone, and this is all me asking ChatGPT a few questions. What it came back with is no surprise to me.
The New AI Reality: Big Earnings, Bigger Data Centers — Is It Becoming a Ponzi Scheme?
November 16, 2026 | AI Investment and Market Risk
The AI boom is real, the revenues are real, and the technology is real. But the financing behind the massive AI buildout is starting to look increasingly circular. That is where the danger lies.
I would not call the entire AI industry a Ponzi scheme. A Ponzi scheme involves fraud, where money from new investors is used to pay earlier investors while concealing the absence of legitimate returns.
What we're seeing in AI is something different: companies investing in one another, borrowing billions, signing enormous long-term contracts, building data centers, and purchasing chips from companies that are also helping finance the expansion.
That can be perfectly legal. It can also create a financial bubble if actual customer demand fails to justify the spending.
1. The money being spent is staggering
2026 PLANNED CAPITAL EXPENDITURES
$720B–$740B
Amazon, Alphabet, Microsoft and Meta combined
$730 billion midpoint
Approximately 78% above their combined 2025 capital spending, based on comparable estimates.
Company guidance and analyst compilations; includes non-AI infrastructure spending.
November 2026
Big Tech Capital Spending Plans — 2026
USD billions, approximate guidance; includes spending not exclusively attributable to AI.

That is $730 billion from just four companies, before including Oracle, CoreWeave, dedicated AI startups, and other data-center developers.
The four major companies had already spent approximately $508 billion during the first nine months of 2026, according to an industry compilation.
Report AI
2. The circular financing problem
Here's how the cycle can work:

Circular commercial relationships are not automatically fraudulent. But they make it harder to distinguish demand from independent customers versus demand supported by financing within the same ecosystem.
Financial analysts have identified these interconnected investments and contracts as an important risk in the AI infrastructure market, https://gfmag.com/technology/the-circle-game/
3. The biggest warning signs
Massive financial commitments
Anthropic's November 2026 SEC filing disclosed approximately $900.9 billion in compute and infrastructure commitments, extending over many years. Roughly $70.7 billion is due over the 12 months following September 30, 2026 https://www.sec.gov/Archives/edgar/data/2130357/000162828026073045/anthropicpbc-fwpjapanamend.htm
Borrowing against future demand
AI infrastructure companies increasingly use debt, leases, project financing, and long-term customer contracts to fund expansion. If growth slows, debt obligations remain even when computing capacity sits unused.
Expensive chips that depreciate rapidly
A $10 billion investment in GPU infrastructure today could face competition from substantially more efficient chips within a few years. The equipment may still work, but its earning power could decline faster than investors anticipate.
The spending is running ahead of cash generation
During the third quarter of 2026, several major technology companies reported capital expenditure exceeding their quarterly operating cash flow. That does not necessarily indicate financial distress, but it increases the importance of future returns. https://www.thecapex.report/hyperscaler-capex
4. But AI is generating real revenue
This is the strongest argument against calling the boom a Ponzi scheme.
A recent Federal Reserve research paper estimated that the combined annualized revenue of major frontier AI laboratories was at least $107 billion in Q3 2026, based largely on public company disclosures. The paper cautioned that much of this information was unaudited and could not be treated as fully reliable.
That revenue is substantial, but it is still small relative to the infrastructure investment and contractual obligations being accumulated across the industry. The figures are not directly comparable because they cover different companies, periods, and accounting measures.
5. What could cause an AI bubble to burst?
The biggest risk is not necessarily that AI stops growing. It is that AI continues growing, but not fast enough to pay for everything being built.
If customer demand disappoints, the consequences could include declining GPU rental prices, excess data-center capacity, lower margins, tighter financing, project cancellations, and pressure on highly leveraged companies.
A slowdown in spending by a few major AI buyers could also affect suppliers across the industry, including Nvidia, networking companies, memory-chip manufacturers, utilities, and construction firms.
For ES and NQ, that matters because the AI trade is deeply connected to technology valuations and earnings expectations.
My bottom line — The Opening Print
I don't think AI is a giant Ponzi scheme, but I do think we're entering a dangerous phase where the amount of money being spent is growing faster than investors can confidently measure the return on that investment.
We're talking about hundreds of billions of dollars being spent on chips, power plants, data centers, cooling systems and networking equipment. At the same time, the companies financing these projects are often investing in one another and signing massive multiyear agreements.
The question is no longer whether AI is going to change the world. It already is. The question is whether the companies spending hundreds of billions of dollars can generate enough profits and cash flow to justify the investment.
And there is one more thing traders need to understand: AI can be the biggest technological breakthrough of our lifetime and still produce one of the biggest stock market bubbles of our lifetime.
We saw something similar during the dot-com era. The internet wasn't a scam, but that didn't stop many internet stocks from losing 80% to 90% of their value when expectations got ahead of reality.
For the ES and NQ, that's the risk. It doesn't take the end of the AI revolution to produce a major correction. All it takes is earnings and future cash-flow expectations falling short of what investors have already priced. As NVIDIA (the grandfather of AI) since 10/7/26:
Based on the 29 trading sessions provided (from October 7, 2026, through November 16, 2026), here is the breakdown of the performance:
Summary Statistics
Total Days Up: 8 sessions
Total Days Down: 21 sessions
Total Net Point Change: -83.31 points (calculated from the starting open of 237.68 on Oct 7 to the ending close of 156.37 on Nov 16)
Sum of Daily Point Changes: -82.87 points
Cumulative Percentage Change: -34.21% (from the 10/07/2026 open of 237.68 to the 11/16/2026 close of 156.37)
Sum of Daily Percentage Changes: -41.03%
Join Carolyn Boroden, the Queen of Fibs, Tuesday, November 17 at 2:00 p.m. ET in Trade Anchors and The Pit for an introduction to advanced Fibonacci time and price analysis. Watch link coming soon.
// OUR LEAN | — |
// MOC | — |

November 16 MOC — selected historical snapshot. View full-size image
MOC Reveal Buying Gives Way to a Net Sell Finish
The Reveal for the MOC on November 16, 2026 was $4.8 billion to buy. Outstanding buys were $7.5 billion against $2.7 billion in sells. The net balance narrowed to $3.2 billion to buy at 3:55 p.m., then crossed to $0.5 billion to sell at 3:56. Outstanding sells increased by $0.6 billion across that minute, while buys contracted by $3.1 billion. The final update showed $1.9 billion to sell. The 3:55 p.m. period through the close is usually discretionary traders closing out the day; these snapshots do not identify participants. Both sides subsequently contracted, consistent with normal pairing, without proving its precise cause.
At the Reveal, buyers concentrated in technology: Apple carried $625.6 million, Microsoft $481.9 million and Nvidia $407.9 million in buy imbalances. AMD led sellers at $135.9 million, followed by Meta at $98.8 million. That split shows disagreement within growth stocks. These are individual outstanding imbalances, separate from the paired-dollar column and market totals.
The Reveal sector mix reinforced that concentration. Information Technology showed $2.4 billion net buying, with 63.4% of symbols buying. Consumer Discretionary added $652.8 million. Utilities leaned $185.5 million to sell. Health Care had net buying despite a majority of symbols selling. These describe the initial mix, not a verified final-close sector breakdown.
Overall, the buy-heavy opening indication did not survive the closing sequence. The turn in net pressure warrants attention, but neither imbalance contraction nor the final feed update proves a price move or auction execution. Historical size rankings and day-to-day comparisons remain unestablished because the prior comparison cohort carries an unresolved source conflict.
Want to follow the auction pressure behind this recap? Explore the MIM’s live imbalance snapshots and sector view.

Four-market leans at the selected historical snapshot. View full-size image

November 16 MOC imbalance sequence, millions of dollars. View full-size image

November 16 MOC minute snapshots through 16:00. View full-size image

Top buy imbalances at the selected historical snapshot. View full-size image

Top sell imbalances at the selected historical snapshot. View full-size image

Sector imbalances at the selected historical snapshot. View full-size image
// PIT TRADING ROOM | — |
Don’t trade the open alone. Join us live in The Pit for free, follow the conversation, and bring your questions.
Tuesday, November 17: Quick Draw Bo / Bosier Parsons joins The Pit at 9:15 a.m. ET. Scheduled YouTube link pending verification.
// LEVELS | — |
No BTS levels today. Our data provider was knocked offline yesterday.
// CALENDAR | — |
Economic Calendar

Earnings Calendar

// BREADTH DATA | — |
Completed U.S. cash session; reported closing TRIN when available. Source: WSJ Markets Diary. Feed: Monday, November 16, 2026. | ||||||||||||||||||||||||
WSJ Weekly Totals; TRIN calculated from weekly issues and volume. Highs/lows are not unique-name counts. Source: WSJ Markets Diary. Feed: Friday, November 13, 2026. | ||||||||||||||||||||||||
// IN OUR TRADING ROOMS | — |
Polaris Trading Group
Polaris Trading Group’s Monday, November 16 session began with a Cycle Day 3 recovery objective and ended with David describing a market offering little shelter for the bulls. The room worked through an overnight reversal, changing intraday reference levels and a brief options bounce before attention turned to protecting gains and managing the final hour.
Overnight setup and the cycle objective
David marked the line in the sand at 7595. His early recap said the overnight move had first fulfilled the 7615, plus or minus five points, target zone before reversing and reaching the lower 7545 target.
By the morning update, price was trading below the Cycle Day 1 low at 7572. Reclaiming that level was the cycle’s main objective, rather than an accomplished recovery.
The overnight targets and the later reclaim objective described different stages of the plan. Reaching the earlier zones did not establish that buyers had repaired the subsequent weakness.
Morning conditions and nearby resistance
David identified a put wall at 7520 and subsequently reported a 20-point bounce from that area. That response gave the room a nearby reference while the larger recovery objective remained overhead.
After the cash open, he said bulls needed to clear and hold the 7540 handle cleanly for the potential to move higher.
As the morning developed, David marked a 7495 one-sigma reference and then described the active trading sandbox as 7475–7500. Those were the immediate working levels in the later discussion.
Options opportunity and reported pricing
A question about a possible same-day options play followed the discussion of a shift toward the Cycle Day 1 low. David responded that it was worth a play.
In subsequent updates, he said the 750 call had traded around 0.30 when the question was raised and was then trading around 0.77. He separately identified the SPY Cycle Day 1 low with the 750 strike.
Those posts describe David’s observed option pricing. The room’s discussion did not establish a common entry or exit for everyone following the idea.
Risk management as the session weakened
Later morning discussion warned that failure to hold a nearby level could bring another test of the lows. The tone remained conditional, with attention on whether an attempted bounce could hold.
By early afternoon, David was reporting sharp pressure in broad equities and technology alongside a jump in volatility. His description shifted toward indiscriminate selling.
The gap between the early reclaim objective and the afternoon tone mattered: the room continued to discuss the recovery level, but the later posts did not confirm that it had been recovered.
The final hour and closing imbalance
David’s late-session comments remained defensive, including a warning that there was no hiding for the bulls. Around 3:30 p.m. ET, he urged traders to monetize their same-day options as time ran down.
Near the close, he posted a reported $5.1 billion MOC buy imbalance and initially described it as support for the bulls. A later update said that buying had been easily absorbed.
The closing sequence captured the day’s tension: an apparent source of buying interest arrived, yet David’s subsequent observation remained cautious about how much relief it provided.
What carried through the day
The room moved from an overnight target map to a reclaim test, then to tighter intraday levels and late-day options management. Keeping those stages separate preserves the useful trading sequence.
The practical emphasis was on responding to price at the working levels and managing a short-lived opportunity as conditions changed, rather than treating the morning recovery objective as a guaranteed destination.
