“Tech Bubble” Warnings Cost Investors a 550% Nasdaq-100 Run

Investors have been hearing “tech bubble” warnings for more than a decade — but instead of collapsing, the Nasdaq‑100 has gained 550%. If we look back ten years ago to 2015, headlines such as “Sell everything! 2016 will be a cataclysmic year” confronted investors with calls for an imminent recession. The bears made repeated claims that a “tech bubble” was about to burst with some of the world’s most prominent venture capitalists drawing parallels to the dot-com era. 

What followed tells a very different story with not only the Nasdaq-100 up 550% over a 10-year period but also high-flying stocks like Shopify returned as much as 5200% and Nvidia returned 22,000% over the same period. 

It’s true that capturing those gains does not come easy. Investors had to hold through five drawdowns that were greater than 20%, including two declines greater than 30%, while tuning out a constant stream of bearish commentary – often from reputable sources – proclaiming the long-awaited tech bubble has finally “popped.” Despite these strong convictions, the long-term trend remained intact. 

Below, I examine periods when normal market resets were mischaracterized as a bubble. I then discuss why today’s AI cycle does not share those characteristics before concluding with the technical signals we are monitoring to confirm that AI remains in a sustained uptrend. 

Timeline of Tech Bubble Talk:

The idea of a “tech bubble” — now rebranded as an “AI bubble” — is not a unique prediction.   

• 2015: Headlines warned that tech valuations were unsustainable, with calls to “sell everything” ahead of a widely anticipated 2016 recession. Particularly, there were concerns due to high valuation of Chinese tech stocks. “Chinese technology stocks do resemble the dot-com bubble,” Vincent Chan, the Hong Kong-based head of China research at Credit Suisse Group AG, said in an interview on April 2. “Given stocks fell 50 to 70 percent when that bubble burst in 2000, these small-cap Chinese shares may face big corrections when this one deflates. On the other hand, the US was also preparing to raise interest rates for the first time since the financial crisis of 2008.
• In 2015, despite the warnings, broader U.S. markets proved more resilient than feared, with the Nasdaq-100 rising 8.4% in 2015. Even though these narratives continued in 2016, the Nasdaq-100 managed to close in green with a gain of 5.9% in 2016.   
• In 2017, investors were once again worried about high valuations and the unwinding of quantitative easing. Despite all the concerns, the Nasdaq-100 rose 31.5% in 2017.   
• In 2020, despite rapid earnings growth, with companies like Zoom growing 326%, tech stocks were labeled a bubble amid pandemic-driven volatility and policy uncertainty. “Everybody loves a party … but, inevitably, after a big party there’s a hangover,” the billionaire investor Stanley Druckenmiller said in a Squawk Box interview. “Right now, we’re in an absolute raging mania.”” Although tech would later reset, the most explosive move followed these bubble warnings. 
• In 2022, rising rates and tightening financial conditions reignited claims that the tech bubble had finally burst and the Nasdaq-100 was down (33%) in 2022. However, tech stocks recovered in 2023, and the Nasdaq-100 rose 53.8% in 2023.  
• In 2024, the narrative re-emerged once again—this time framed as an “AI bubble”—despite strong balance sheets, accelerating earnings, and durable long-term demand drivers. However, Nasdaq-100 rose 24.9% with Nvidia rising 171.2% in 2024. 

Supply Constraints Make a Widespread AI Bubble Unlikely

Bubbles are typically defined by oversupply, yet many pockets of today’s market are the opposite – they are supply constrained.  

During the dot-com era, the market was flooded with far more e-commerce and internet sites than demand could support, largely because barriers to entry were low. AI is quite the opposite as it’s an expensive technology that has a very high barrier to entry and lacks democratization. The buildout is constrained across critical inputs, such as compute, memory, networking, power, and advanced packaging, which further raises the hurdle for new entrants and slows supply growth. 

TSMC, the world’s largest contract chipmaker and critical supplier for AI logic chips, has stated that its advanced-node capacity is roughly three times short of what AI demand requires, even amid ongoing expansion efforts. This reflects a real gap between what customers want and what TSMC can physically produce.  

AI data centers and accelerators consume an outsized share of DRAM and HBM, driving a global memory shortage that is now described as unprecedented and likely to persist beyond 2026. Companies like Tesla and Apple have also signaled a shortage of DRAM. The shortage has also led to the price hikes, and the cost of one type of DRAM soared by 75% from December to January. 

Energy is another gating factor that makes an “oversupply” outcome unlikely anytime soon. In 2024, we highlighted Wells Fargo’s projection that AI power demand could surge 550% by 2026, rising from 8 TWh in 2024 to 52 TWh, before accelerating another 1,150% to 652 TWh by 2030 — an 8,050% increase versus the 2024 baseline. Under that framework, training drives the bulk of demand earlier in the cycle, while inference becomes a larger driver later in the decade. 

There is more data from IEA, which projects global data center electricity demand will more than doube from ~415 TWh in 2024 to ~945 TWh by 2030 in its base-case scenario. Under the agency’s AI “lift-off” scenario, demand reaches 1,250 TWh — a trajectory that more closely aligns with Wells Fargo’s more aggressive outlook. 

The message from management teams is consistent with these forecasts: Google CEO Sundar Pichai noted that Google has been “supply constrained” even as it ramps capacity, citing longer supply-chain time horizons. Meta CFO Susan Li echoed the same point, saying Meta remains capacity constrained and will “likely still be constrained through much of 2026” until additional capacity from its own facilities comes online. 

AI is Driving Significant Revenue and Profits

The late 1990s was defined by pre-revenue companies with many dot-com darlings reporting a mere $10 million to $20 million in revenue. This fact alone has been the reason other so-called tech bubbles were rather quick corrections such as mobile, social media, cloud infrastructure, and now AI are all trends that drove significant revenue and profits for public companies.  

Our firm was early to point out that Meta was now second to Nvidia on AI revenue with its AI ads automation platform reaching a $60 billion run rate in 2025; three-and-a-half years from launch. launch. Similarly, we can look at OpenAI’s trajectory from $1 billion in revenue in 2023 to an estimated $20 billion annualized revenue in 2025 – which represents the steepest rise in technology history.  

When you compare a successful dot-com company, the profile is very different from what we see in AI stocks today. Amazon rose 3,320% from January 1995 to March 2020 yet the bottom-line was deep in the red with a negative profit margin of (56%).  Meanwhile, a few AI companies like AppLovin, TSMC, and Reddit have reported net profit margins of 66%, 48%, and 35%, respectively in their recent Q4 results; a stark contrast to one of the dotcom bubble’s best stocks.  

Notably, prices across leading AI companies are not broadly untethered from fundamentals. In fact, a handful of companies are fundamentally cheaper now than they were at the first sight of this AI-driven boom in early 2023. For example, Nvidia currently trades at 21.7x forward earnings, and is lower than the multiple it traded the day prior to its May 2023 Hopper-driven blowout earnings report, despite shares rising 475% over the same period. 

Source: YCharts

Most importantly, those funding AI are highly profitable tech companies compared to venture capitalists who pushed unprofitable companies public for a quick exit during the dotcom bubble. Big tech is not looking for a quick exit and on track to spend $655 billion in 2026, up 60% YoY. 

AI Is Early Cycle; Not Late Cycle

Cycle timing is arguably the most critical part of being a tech investor. Enter too early, with autonomous vehicles being a good example, and the opportunity cost can be very high as capital sits idle while adoption lags. Enter too late, like chasing the dot-com boom at the peak, and you risk buying the top with little hope for a near-term recovery. Equally as painful is closing a promising stock too early in the cycle, and seeing it rise sharply in the years that follow.  

Because cycle timing is emotionally taxing, investors often equate sharp downside volatility with “bubble” conditions—yet the two are not the same. 

The smartphone cycle was one of the most powerful product cycles in modern history. Apple launched its first iPhone in June 2007, just four months before the market topped in October. When the market top occurred, smartphone adoption was still in its nascent stages, leaving little opportunity for bubble dynamics to form; exiting the GFC, smartphone adoption proved to be robust, with TTM growth of 63% from August 2008 to August 2009, while also marking one of the fastest 10-quarter adoption curves in consumer tech history. Momentum on the app side was explosive with the App Store reaching 1 billion paid and unpaid downloads within nine months.   

Despite the rapid adoption post-iPhone launch, this did not insulate Apple from realizing several meaningful periods of volatility, including a 61% drawdown shortly after that launch in the 2008 bear market. Over the next decade, Apple again faced two major drawdowns of 45% and 34%; however, shares ended this decade more than 724% higher, highlighting that extreme volatility does not always mean it is a bubble. 

Salesforce is another stocks that saw shares enter a multi-month drawdown of 73% in late 2008, though shares quickly returned to new highs in just 16 months and went on to rally 131% in the time it took the broader market to exit from the bear market. This is because Salesforce was still witnessing rapid revenue growth early in its adoption curve – revenues more than doubled to over $1 billion in the two years from 2007 to 2009, and by 2013, revenue had surpassed $3 billion. 

Looking back to a similar time period, AWS is an excellent example of the build phase versus the yield phase to where it required extensive upfront capital that later became fast growing revenue. AWS revenue grew by 24% YoY in Q4, accelerating 4 percentage points from 20% growth in the previous quarter and was the fastest growth in the last 13 quarters.  

Cloud Software Nearing the End of its Cycle

The cloud category has treated investors quite well with recurring revenue, resiliency during Covid, and some of the strongest examples of product-market fit available on the public markets over the past ten to fifteen years. However, I made the argument three years ago that cloud software was in the later innings of its cycle, as many best-of-breed companies saw growth fall off a cliff in 2022 and 2023.  

We had pointed out in our free analysis in late 2022, Slowing Growth In Cloud Stocks: When Will We Hit A Bottom, that nearly all cloud companies were reporting a notable, sequential slowdown between Q3 to Q4. These Q4 2022 guides marked a ‘historic slowdown’ for the once-resilient category, as Q4 is typically the strongest seasonal quarter for the industry. Snowflake was a prime example of this, as it had guided for 3% QoQ growth in Q4 2022, what would mark a 12 point decline from 15% QoQ the year prior. 

We followed up in March 2023, asserting in the analysis, Slowdown In Cloud Stocks On Thin Ice Following Q1 Guides, that hyperscalers were seeing growth rates plummet. AWS reported Q4 2022 growth of 20%, half of the 40% reported in Q4 2021 while guiding for Q1 2023 growth in the mid-teens; Azure saw a similar nearly 20 point deceleration from 49% in the March 2021 quarter to 30-31% guided for March 2022.  

These decelerations are clearly visible looking at some of the best-of-breed names over the last five years. Snowflake reported revenue growth north of 100% in Q1 2022, yet exited 2023 seventy points slower at 31% YoY; Twilio decelerated from 67% to the low single-digits.  

Source: YCharts

Simply put, the hypergrowth cloud era from 2015 through 2021 has passed, with sharp growth deceleration as rates rose leading to multiple compression. Those who think AI is disrupting cloud software are not accounting for the fact that cloud software was as the end of its cycle and ripe for both consolidation and disruption (a terrible place for investors to be positioned) – which is why I went to great lengths to make sure my premium research members knew to steer clear of cloud software three years ago.  

There is also more evidence that cloud is now late cycle, with the market being extremely saturated with more than 30,800 SaaS companies worldwide, each competing with one another for wallet share as production differentiation narrows. Market saturation preceded the disruptive fears from AI-based solutions automating workflows.

Today, the same narrative has resurfaced around AI. This analysis breaks down why the AI market is fundamentally different from past bubbles, why corrections have been misinterpreted, and what indicators we’re watching to confirm the trend remains intact. 

There Will Be a Correction; It Won’t Be a Bubble

While we do not believe AI is a classic bubble, that doesn’t mean we won’t see meaningful selloffs that create attractive buying opportunities. In fact, warning signs have been building since October 2025 that we may be approaching one of these pullbacks. One of the more concerning signals is that the Magnificent 7—what I consider the generals of this market—appear to have topped well before the S&P 500. 

Since November 2021, periods in which the equal-weight Mag 7 index fails to confirm new highs in the S&P 500 have been a reliable signal of a weakening market environment. A similar divergence is developing today, and until it resolves to the upside, it remains a meaningful warning for the durability of the broader uptrend. 

Looking under the hood, all 7 of the Mag-7 are currently making lower highs while the S&P 500 made higher highs. For reference, this is roughly 32% of the S&P 500’s weight that is preventing the broader market from moving higher.  

Also, before volatility started picking up, we noted in prior reports, both retail and professional investor sentiment are elevated to historically concerning levels, suggesting an environment where risk is being discounted and investors behave as if there is no price too high. 

The AAII weekly survey (retail sentiment and positioning) and the NAAIM weekly survey (professional manager exposure) are at levels that exceed the average for most major tops. Since late October—around when several markets began topping out—NAAIM readings have ranged between the 78th and 96th percentile of all bullish readings, suggesting managers have been heavily allocated to equities for more than three months, and maintain this exposure.  

When compared to levels seen before prior market tops, these readings suggest sentiment and positioning are among the more extreme observations on record. 

In other words, both retail and professional investors appear to expect higher prices and have expressed that view through high equity exposure. What is more concerning is that margin debt in the U.S. is at record highs, surpassing the 2021 peak.

These conditions often precede periods of volatility. However, late-cycle behavior does not automatically imply that a bubble is forming. Since 1980, only a small number of major market peaks coincided with the bursting of true systemic bubbles—most notably technology in 2000 and housing and credit in 2007. Many other peaks resolved into corrections and recoveries within longer secular bull markets. 

In terms of where this volatility could take us, the below scenarios are the most probable based on the current price action: 

• Green – If we can continue to bounce through SPX 6869, 6901 and finally 6952.50, then we will likely push toward the 7200 range in the coming weeks. This will complete the final 5th wave in a very extended uptrend that started off the April low in 2025. If this happens, we will look for more stocks and markets to not make new highs with the S&P 500 to further confirm we are setting up for a period of volatility.
• Red – In our last broad market article we outlined the importance of the 6780 – 7720 region in SPX. So far, these levels have held and it is where the market staged a bounce.

  These same levels remain of utmost importance for the bulls. If they break, then the period of volatility has already begun as we head toward 6500 – 6300 in the coming weeks. This will likely complete the first leg in a larger correction, as we mount a bounce that makes a lower high into later 2026. 

Conclusion:

AI will almost certainly deliver more volatility, and investors should expect meaningful selloffs that create buying opportunities. But volatility is not proof of a classic bubble. The dot-com era was defined by oversupply and fragile fundamentals; today’s AI buildout is being led by the world’s strongest operators, backed by real revenues and profits, and constrained by hard limits in compute, memory, networking, and power. 

The more important question isn’t whether we’ll see a pullback — it’s where we are in the cycle. AI is still transitioning from the training phase into the inference phase, where monetization will accelerate and the “capex with no revenue” narrative will begins to fade. In other words, the loudest bubble debates are arriving before the most important revenue engine fully turns on. 

We’ll continue to watch the same signals that matter in every tech cycle: whether fundamentals keep compounding, whether supply constraints remain binding, and whether the market’s leadership confirms a durable uptrend. If those conditions hold, then “it’s a bubble” may once again prove to be the most expensive words in tech for those sitting on the sidelines. 

Since our inception in May 2020, I/O Fund has delivered a cumulative return of 326%— if we were a hedge fund, we’d rank #1 and if we were a tech ETF or Mutual Fund, we’d rank #3 in the United States.   

Being early to many lesser-known AI winners helped us to achieve these results. To get our Top 15 AI stocks, real-time trade alerts, weekly webinars and deep-dive research from a proven team in AI and tech stocks, Sign up now.

Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.

Recommended Reading:

• My Top 2026 Stock Pick for the AI Boom
• I/O Fund Jumps to 326% Cumulative Return, Ranking Among Wall Street’s Best
• Bitcoin After the Cycle Peak: What Comes Next and How We’re Positioning
• S&P 500 Outlook 2026: Rising Volatility Risk and Key Support Levels