1999-2000 Tech Bubble vs. Ai Boom 

One of the things I have always said is that no one can predict the future, but like the tech bubble, there were a lot of pumped up stocks that didn’t have the money to stay in the business. There are some striking similarities between the dot-com bubble of 2000 and today's AI infrastructure boom, particularly the enormous spending, dependence on outside capital, rising debt, and uncertainty about future profits. The biggest difference is that today's leading AI investors include some of the most profitable companies in the world.

The Biggest Similarities

I know this is not the same as it was in 2000, but the Nasdaq Composite has rallied approximately 131% since ChatGPT launched in November 2022, a striking similarity to the 130% surge during the 1999–2000 dot-com bubble, when the index gained another 24% from December 31, 1999, to its March 2000 peak before collapsing nearly 78% by October 2002.

Is AI Running Into A Financing Problem?

Well, the ES and NQ charged higher for the second day after Hock E. Tan, the President and CEO of Broadcom ($AVGO), held weekend meetings with anxious executives from Bank of America, Citigroup, and Morgan Stanley to get the banks to commit more than $40 billion to the largest artificial intelligence debt deal ever. The financing is intended to allow Anthropic to lease specialized chips that Broadcom is developing with Google, ahead of Anthropic's highly anticipated initial public offering. Basically, the banks received stronger guarantees from Broadcom to protect them against losses if Anthropic defaults on the loans. I think that's the good side of the story, but the bad side is that Meta ($META) announced it was looking to raise money. Reuters reported on October 19 that OpenAI was approaching investors, including SoftBank and Abu Dhabi's MGX, ahead of Anthropic's expected IPO. Bloomberg separately reported that Broadcom was in preliminary discussions about another $30 billion financing arrangement for OpenAI. CoreWeave ($CRWV) has approximately $50 billion in existing debt supporting its AI infrastructure expansion, while Oracle ($ORCL) is reportedly in discussions with lenders to secure additional financing for AI chips.

But it's not just the chipmakers. SpaceX/xAI is reportedly in talks with banks, Apollo, and PIMCO to raise $40 billion to finance Nvidia chips, while China's DeepSeek is reportedly nearing completion of a $15 billion fundraising round backed by Chinese investors. Meanwhile, AI infrastructure company Firmus has shifted to private fundraising after shelving its planned $5 billion IPO, AirTrunk has postponed a proposed $1.5 billion Singapore REIT offering, and Databricks previously explored raising capital at a valuation of $165 billion to $175 billion, although its current fundraising status remains unconfirmed.

How Big Are Your Pockets 

On April 30, Amazon, Alphabet, Microsoft, Meta, and Tesla were going to spend as much as $750 billion on capital expenditures in 2026, but by September 25, Goldman Sachs had raised its estimate for leading AI hyperscalers to approximately $800 billion in 2026 and projected spending could reach $1.2 trillion in 2027. The biggest concern for Wall Street is that the five largest AI hyperscalers are increasingly relying on debt and outside financing to fund their massive infrastructure investments, with Reuters reporting that their combined capital spending could exceed their total free cash flow by 2027, raising concerns about growing debt, higher borrowing costs, and whether AI investments will generate enough returns to justify the spending.

Conclusion: The Good and the Bad of the AI Infrastructure Boom

The positive side is that the AI infrastructure boom represents one of the largest investment cycles in history, creating enormous opportunities for semiconductor companies, cloud providers, data centers, energy producers, utilities, construction companies, and the broader economy. The demand for computing power continues to grow, and companies like Amazon, Microsoft, Alphabet, Meta, and Oracle are investing hundreds of billions of dollars in infrastructure they believe will drive productivity, innovation, and future earnings growth. If AI adoption and revenues continue to accelerate, these investments could generate significant long-term returns.

The negative side is the staggering cost of building all this infrastructure and the growing dependence on debt and outside financing. With the five largest hyperscalers expected to spend approximately $800 billion in 2026, the question is whether AI revenues can grow fast enough to justify the spending. Higher borrowing costs, rising debt levels, pressure on free cash flow, power shortages, and the possibility of excess data-center capacity are legitimate concerns. There is also the risk that companies are spending aggressively to avoid falling behind their competitors without knowing when those investments will become profitable.

My takeaway: I don't question that AI is going to change the world, but that doesn't mean every company spending billions of dollars on it will make money. We have seen this before with the internet and telecom booms: the technology changed the world, but many investors still lost fortunes. The AI buildout could continue supporting the stock market for years, but if spending keeps rising faster than revenues and free cash flow, Wall Street will eventually demand results. With so much of the S&P 500 and Nasdaq concentrated in companies tied to AI, any disappointment in earnings, returns on investment, or financing conditions could have consequences well beyond the technology sector. The question isn't whether AI is the future. It's whether Wall Street has already priced in too much of that future.

What we know is that when there is good news in the tech and AI space, the markets react in kind, and then you throw in oil falling and when the bonds are firm. It is as they say, a perfect storm, especially when the ES and NQ are oversold and the big accounts are hedged. But there are always buts. I think AI companies have underestimated their infrastructure costs. One of my biggest concerns about the AI boom is whether these companies have fully accounted for what it will cost to build and maintain the infrastructure needed to support it. The five major hyperscalers are spending hundreds of billions of dollars on chips, data centers, electricity, and networking equipment, but there is still uncertainty about how quickly those investments will generate adequate returns. Reuters reported that their combined capital expenditures could exceed their free cash flow by 2027. What happens when the cost of building AI infrastructure continues to rise, the chips need replacing, borrowing costs remain elevated, and revenues don't grow as quickly as expected? I am not saying AI is a bubble, but there is a big difference between developing technology that changes the world and making enough money to justify what you spent building it. At some point, Wall Street is going to want to see the return on investment, not just another announcement about how many billions are being spent.

PTG Cycle Day 2 — Balance and Consolidation

By David D Dube, Polaris Trading Group

S&P 500 (ES)

NEW PTG Trading Room Recordings

We are now recording the PTG Trading Room Morning Session. These will be “raw” unedited and possibly lengthy. While watching, adjusting the playback speed is recommended. You will be able to find the most recent five (5) session recordings here: Polaris Trading Group Videos

Note: Trading Room RECAP archives link: PTG-RECAP

***PTG would appreciate you taking a moment to leave some feedback at Trustpilot Reviews regarding how the Daily Trade Strategy (DTS) is helping your trading. Your input helps us continue refining the edge and delivering greater value to our community.

🎯 Cycle Day 2 Expectations

🔄 The balance process begins
🎯 Range development starts to take shape
⚖️ Buyers and sellers square off for control
💥 Reversal potential quietly increases

In other words…

Cycle Day 1 established the key low @ 7700
Cycle Day 2 installs the shock absorbers.

This is where the market says:
“Alright… everybody calm down… let’s figure this thing out.”

And for PTG traders?
That’s your cue to shift gears:

✔️ Less emotion
✔️ More structure
✔️ Cleaner, more deliberate opportunities

No need to force trades.
No need to chase moves.

Just stay patient… stay disciplined… and let the market come to you.

⚖️ Cycle Day 2 Objective: Balance… Consolidate… 

Stay Patient.
Stay Disciplined.
Stay PTG.

🎯 Scenarios in Play

🟢 Bull Case — Buyers Stabilize & Reclaim

Acceptance Above: 7750 ±5

Upside Objectives

• 7765

• 7775

• 7785

This signals responsive buying evolving into initiative control.

⚠️ But remember:
This is recovery mode — not dominance yet.

🔴 Bear Case — Continued Rotation / Controlled Reset

Acceptance Below: 7750 ±5

Downside Objectives

• 7735

• 7725

• 7715

This is not panic selling —
This is orderly distribution… the kind that grinds traders down.

📊 Key Reference Levels

• PVA High Edge: 7772

• PVA Low Edge: 7743

• Prior POC: 7758

• Cycle Day 1 Low: 7700

👉 Important:
These levels cluster tightly — forming a decision zone, not noise.

⚠️ Tactical Takeaway

Of course, nothing changes for PTG…Simply follow your plan. Take only Triple A setups and manage the $risk. ALWAYS HAVE HARD STOP-LOSSES in-place on the exchange.

PTG’s Primary Directive (PD) is to ALWAYS STAY IN ALIGNMENT with the DOMINANT FORCE.

   ES

“The safest place to look for a new trade is at the end of the first correction to a new swing.” — W. D. Gann

David D Dube
Polaris Trading Group

About David Dube

David D Dube has been a professional trader for over forty-one years, including senior trader management roles at Lehman Brothers and UBS Investment Bank. He brings deep global markets experience to the PTG community.

Read about PTG’s Daily Trade Strategy Briefing.

Disclaimer: Trading of securities, options and futures may not be suitable for everyone and involves the risk of losing part or all of your money. Commentaries are educational in nature and are designed to contribute to your general understanding of financial markets and technical analysis. Use it how you want and at your own risk. We are not registered investment or commodity trading advisors. This information is a general publication that reflects our own opinions and is not a specific recommendation to any one individual. You must consult your own broker or investment advisor for investment advice. Past performance is not a guarantee of future success. Controlling risk through the use of protective stops is essential.

Tuesday’s ES session extended the rebound, but buyers had to work through another overnight dip before taking control. The session opened at 7715.00 and reached an overnight low of 7699.75 around 4:00 a.m. ET. That test held, and the recovery carried through the 7715 area that had been central to the prior day’s close. The overnight high reached 7749.25, putting the market back near 7750 before regular trading began.

By the cash open, the ES was at 7734.75, up 20.25 points from Monday’s settlement. The overnight advance did not eliminate the risk of an early shakeout: price still had to establish acceptance above the prior value area. Holding the overnight low left buyers with a defined reference, while the recovery toward 7750 showed that the dip had attracted demand. The overnight session finished with 324,812 in volume and a 19.75-point gain from its opening price.

Regular trading opened at 7734.75, then pulled back to 7712.50 before buyers rebuilt the advance. That early low stayed above the overnight extreme, and the market subsequently pushed through 7750. The afternoon carried the ES to its 7772.75 session high during the 3:45 p.m. ET interval. At 3:50 p.m., the last traded price was 7765.75, leaving the market near the upper end of the day’s range before the closing rotation.

The cash session closed at 7758.25, a gain of 43.75 points, or 0.57%, from Monday’s official settlement. Tuesday’s official settlement was also 7758.25. After the cash close, the ES traded between 7755.25 and 7764.50 and finished the full futures session at 7756.75. Full-session volume was 1.51 million, with 1.13 million during regular trading.

The NQ outperformed, settling at 29812.00, up 374.75 points, or 1.27%, after reaching 29881.25. Its cash-session close was 29813.25, and the final futures-session price was 29803.00. In the end, both indexes carried their recovery into a second session. The ES held its overnight test, cleared the 7750 area, and finished well above Monday’s settlement, although the late pullback left it below the afternoon high. Wednesday’s Fed decision is the next test of whether that recovery can hold.

Technology Selling Meets Health-Care Demand

The Reveal for the MOC on October 27 was $3.3 billion to sell. Gross outstanding selling increased from $7.0 billion to $7.7 billion in the next snapshot, while buying stayed near $3.7 billion. That increase in unmatched selling mattered more than the later contraction. Buying subsequently reached $4.2 billion. The net balance turned $1.0 billion to buy at 3:56 p.m., then returned to $0.3 billion to sell at the final update.

The selected snapshot’s sellers were concentrated in technology: Broadcom showed $576.9 million, Apple $496.8 million and AMD $402.2 million. Micron, Intel and Microsoft added to that pressure. Health-care buyers provided an offset, led by Eli Lilly at $251.8 million and Johnson & Johnson at $183.7 million. Alphabet’s $155.1 million buy showed that technology-linked demand was not uniformly absent.

The sector view reinforced that concentration. Information Technology showed $4.6 billion to sell, against $623.4 million to buy in Health Care. Financials had $272.3 million to buy despite a majority of its symbols leaning sell. At the Reveal, All Markets also had more buying than selling symbols, yet the dollar balance remained firmly negative. These selected-snapshot breakdowns describe the mix, rather than a final-close sector allocation.

Overall, this felt like concentrated technology selling with meaningful offsets, followed by a two-sided closing rotation. Both outstanding sides narrowed into the close, which is consistent with normal pairing and does not establish a directional fade. The 3:55 p.m. period through the close often includes discretionary traders closing out the day, but these readings do not identify participants. A comparable historical ranking is not yet established.

Want to follow the auction pressure behind this recap? Explore the MIM’s live imbalance snapshots and sector view.

How the NYSE and Nasdaq auctions work

How to read the MIM

David Dube / Polaris Trading Group — Live in The Pit

Wednesday, October 28, at 9:15 a.m. ET.

Scheduled YouTube link pending verification.

Don’t trade the open alone. Join us live in The Pit for free, follow the conversation, and bring your questions.

Economic Calendar

Earnings Calendar

Daily Breadth — October 27, 2026

Source: WSJ Markets Diary.

Polaris Trading Group

Polaris Trading Group’s Tuesday discussion began with a constructive Cycle Day 1 setup, shifted defensive after an opening-range failure, and returned to a long lean when buyers defended prior value. David’s updates followed that changing structure, while the room’s conversation kept returning to a practical question: when does a good setup become a poor entry because price has already traveled too far?

Cycle Day 1 and Overnight Structure

• David identified Tuesday, October 27, as the start of a new Cycle Day 1 and cited historical cycle performance of 92.59%. That was the room’s reported cycle statistic, accompanying a conditional plan for the new session.

• His overnight recap described a successful test of 7705 followed by a rally to 7735, matching the targets in the Daily Trade Strategy briefing. Before the cash open, he said price had also come within three ticks of the briefing’s 7750 objective.

The Opening Decision Area

• The initial line in the sand was 7715, with a five-point allowance on either side. Acceptance north of that area favored the bullish sequence of 7725 and 7735; acceptance south favored a rotation toward 7705, 7695, and 7685.

• The prior value-area edges were 7728 and 7705, with the prior point of control at 7714. Those references gave the room specific places to assess whether price was accepting higher value or returning toward the previous session’s center.

Opening Range and the First Shift

• Early in regular trade, David maintained the long theme while price held above the midpoint/VWAP zone. He then reported that the opening range had triggered long plays and anticipated a test of the overnight high.

• That stance changed when price violated the opening-range low: David shifted from neutral to a short lean. The update mattered because it showed the room responding to a failed condition rather than holding the original bullish roadmap regardless of price.

Prior Value Restores the Long Lean

• After a successful test of the prior point of control and the 7715 line in the sand, David shifted back to a long lean. The reference area that had defined the morning plan also became the place where he reassessed the failed opening move.

• As trade developed, he described a double-distribution profile and said bulls would need to power above 7745–7750. That zone was a continuation test, while the prior-value defense supplied the basis for the renewed long lean.

Range Projections and Money Boxes

• Later updates placed price in the Money Box zone. David posted a Cycle Day 1 range projection of 7778.44 and a three-day cycle range target of 7784.85.

• He subsequently reported an A4 runner’s trailing stop being elected, followed by Money Box 2 responding back toward Money Box 1. A pullback to the initial-balance first target at 7761 became his first test of whether bulls would continue buying dips.

• The next update described a Money Box 1 short approaching its DL target. David then said the DLevel Money Box streak continued, preserving the room’s own description of the setup’s progress.

A Setup Is Also About Location

• David presented a buy, sell, or do-nothing question using a chart without revealing how far price had traveled during the session. Participants initially described the structure as bullish, including an upward-flowing group of indicator bands.

• He then changed the question: would they still buy the same setup near the upper end of the session’s range? One participant said remaining room against the ten-day average range could matter and described hesitating to re-enter after price held the overnight high.

The Day’s Practical Lesson

• The sequence moved from long to short and back to long as the opening range and prior value were tested. The useful lesson was the willingness to revise the lean when the condition changed.

• Structure, location and remaining range belonged in the same decision. The room also shared David’s PTG University module on pivots for traders wanting to revisit those reference points.

DTG Room Preview

The DTG group sees a technology-led market heading into an afternoon dominated by the Federal Reserve. Its early preview reports bullish large-trader bias, while emphasizing that the policy decision and press conference could override the morning’s technical setup.

Fed and Technology Leadership

• The key scheduled catalysts are the 2:00 p.m. ET policy announcement and the 2:30 p.m. ET press conference. The group is watching the language on inflation, future tightening, and December rather than assuming the initial reaction will hold.

• AI and semiconductor leadership remain central to the group’s outlook. Microsoft’s earnings and Azure growth are potential tests of the cloud-spending theme; a disappointment could encourage profit-taking in crowded technology positions.

Trend Structure and Decision Area

• The group places the former short-term uptrend-channel bottom at 7741–7746. The new channel bottom is 7665–7670, based on the 7648.50 swing low.

• The group’s 50-day moving-average reference is 7760.25. Bulls need to clear that area to improve the technical picture; the Fed reaction could determine whether that bias changes.

• Its reported five-day average daily range is 82.50 points, down from Monday’s 85.00. The setup still allows substantial movement in either direction.

Trendline References

• Potential resistance zones: 7895–7900, 7970–7975 and 8042–8047.

• Potential support zones: 7741–7746, 7665–7670 and 7424–7429.

• These are the DTG group’s trendline references, with conditional significance. The group cautions that reactions to the Fed can overwhelm the technical framework.