Bull vs. Bear: Is the AI Revolution Nearing a Dot-Com Correction?

Think about how we conceive of AI. Sometimes we use the term interchangeably for tech companies embracing AI. But the AI revolution is about processes, and of course, how new technologies change how business is done.

Valuations may change as companies embrace AI, but no matter what happens with those valuations, in the long term, AI will drive gains for almost every single sector. In the short, medium, and long term, productivity gains will help portfolios significantly.

This broad-based productivity shift is best captured by the ROBO Global Artificial Intelligence Index ETF (THNQ). Unlike narrow bets on single software providers, THNQ tracks the ROBO Global Artificial Intelligence Index, providing diversified exposure across the global AI value chain – from data infrastructure to business processes – capturing the AI hype without the binary risk of a single name.

Industry leaders see AI as a productivity game changer. The most recent comparison markets might make is the internet. The internet helped supercharge global economic value and has contributed massively to global GDP. For the U.S., alone, the digital economy drove some 18% of U.S. GDP.

A.I. could be even bigger. According to T. Rowe Price portfolio manager Dom Rizzo, AI could be the greatest productivity enhancer since electricity. While we want to hold AI stocks, we do so with that as the long term goal, no matter the near term ups and downs.

The $600 Billion Disconnect

Shaw: The long-term productivity story makes sense. AI will likely transform how we work, just like electricity or the internet did. But investors aren’t funding the long term right now – they’re funding 2026, and the math doesn’t add up.

Hyperscaler tech companies plan to spend between $660 billion and $690 billion on AI infrastructure this year alone, according to research from Nick Patience, vice president and practice lead for AI platforms at The Futurum Group. Meanwhile, pure-play AI software companies like OpenAI and Anthropic are projected to generate less than $35 billion in combined revenue.

That’s an almost twenty-to-one ratio. For every dollar of actual AI software revenue, nearly twenty dollars is being spent on the infrastructure to support it. Patience notes that infrastructure built today may take 18 to 36 months to generate proportional returns.

Building that infrastructure requires chips alongside the raw materials and power needed to sustain them. The Amplify Lithium & Battery Technology ETF (BATT) and the Sprott Uranium Miners ETF (URNM) represent the physical side of this $600 billion spend. If the software doesn’t monetize immediately, the demand for the lithium and uranium powering these massive data centers remains a structural reality.

You mentioned the internet comparison. Looking back at the dot-com boom, the internet was real. It did transform everything. But valuations assumed decades of growth would happen in two years. Even winners like Cisco Systems, Inc. (CSCO) took years to recover after infrastructure spending ran ahead of actual usage.

Goldman Sachs data shows markets grew increasingly skeptical throughout 2025. The average stock price correlation across large public AI hyperscalers declined from 80% to just 20% between June and year-end.

The firm notes that investors rotated away from AI infrastructure companies where operating earnings growth is under pressure and where capex is being funded via debt. Although markets aren’t rejecting AI outright, they are differentiating between companies that can actually monetize their spending and those that can’t.

For those wary of the spending race, the Alerian Energy Infrastructure ETF (ENFR) offers a defensive alternative. ENFR provides exposure to the essential energy midstream assets that sustain the massive power requirements of AI build-outs, but with the added stability of a yield-focused, cash-flow-heavy business model that doesn’t rely on software moonshots.

Defining a Bubble

Peters-Golden: So, we’ve established how AI is a long-term play and it touches basically the whole economy. Is it a bubble? The classic vision of a bubble brings to mind Dutch tulips. Feverish investors clamor for assets within a burgeoning sector that seems to deliver almost infinite returns. The dot-com bubble had investors diving into any company that had a dot in the name.

Does the current moment have similarities to those past bubbles? Companies include mentions of AI wherever they can. Companies from Duolingo (DUOL) to cloud company Box (BOX) now claim they are “AI first.” While many are serious about it, it also helps draw the investor eye like the dot-com era.

Let’s look to recent analysis from T. Rowe Price. T. Rowe Price capital markets strategist Tim Murray recently looked at the question of whether AI represents a bubble given comparisons to the dot-com era.

His analysis found that profitability has risen in tandem with those red-hot valuations, diverging from other bubbles. What’s more, he assesses, while investors are right to note the huge capex spending hyperscalers are seeing, most of the AI hyperscalers are still carrying modest overall debt. Credit markets are not yet “sounding alarm bells,” he wrote.

Yes, AI adoption and usage may not be there yet. We need not look far, however, to the world where AI tractors plow fields or AI-empowered robots run factories at all hours of the day.

To capture this “physical AI” shift, the ROBO Global Robotics and Automation Index ETF (ROBO) offers exposure to the global value chain of robotics, providing access to the hardware side of this theme. This means the fund is effectively targeting the industrial application of the technology rather than just the digital hype.

The expansion of AI also has a massive orbital component. The Procure Space ETF (UFO) and the REX Drone ETF (DRNZ) highlight how AI is being deployed in hardware from satellite imagery processing to autonomous unmanned aerial vehicles (UAVs). These are high-utility, industrial applications that look nothing like the consumer hype of the dot-com era.

The space is red hot, and while investors may do well to take some caution, it’s not time to step off the gas entirely.

Where the Profits Actually Are

Shaw: Murray makes a fair point about profitability rising alongside valuations. The hyperscalers are profitable, and their balance sheets can handle the spending. But we need to look at where those profits are actually coming from.

The profitability story works if you’re in the infrastructure business. Nvidia Corp. (NVDA) posted record earnings selling chips to hyperscalers building AI data centers, according to Futurum Research. Those infrastructure companies are profitable too – Amazon Web Services, Microsoft Azure, and Alphabet’s Google Cloud are all generating cash. But the actual AI software companies consuming that infrastructure? That’s a different story.

OpenAI ended 2025 with approximately $20 billion in annual recurring revenue, according to Futurum Research. Anthropic’s revenue run rate surpassed $9 billion in January 2026. Add in smaller players and the entire pure-play AI software sector likely generated around $35 billion in combined revenue for 2026.

Now compare that to infrastructure investment. Tech giants planned to spend roughly $635 billion on data centers, chips, and other AI infrastructure in 2026, according to S&P Global research reported by Reuters. That means for every dollar of AI software revenue, roughly eighteen dollars is being spent on the infrastructure to support it. The build-out is happening years ahead of the revenue it’s supposed to generate.

Murray’s point about modest debt levels also deserves scrutiny. The absolute debt levels may look manageable, but the rate at which capital is being deployed is unprecedented. There’s also the question of whether they can even spend it all. Reuters reported that energy costs and infrastructure capacity are becoming real constraints, with S&P Global’s Melissa Otto noting that persistently high energy prices could force spending revisions.

This infrastructure and data management challenge is precisely why some advisors are looking at the Amplify Transformational Data Sharing ETF (BLOK). While often associated with blockchain, BLOK tracks an index that’s focused on the data-sharing infrastructure required to monetize and secure the massive datasets that AI models rely on, offering a diversified hedge against software-only volatility.

Credit markets aren’t panicking yet. But investors should still think carefully about where they get their AI exposure.

How AI ETFs Can Offer More Specific Exposure to AI Companies

Peters-Golden: Pretty much all equity investors hold some kind of exposure to AI hyperscalers. Those companies like MSFT and NVDA represent big slices of the overall S&P 500. That means that it’s hard to invest “without” exposure to AI, whether its knock on benefits or in semiconductors, the cloud, and other satellite players.

For those who believe this is or is not a bubble with regard to AI, AI ETFs can actually help. Specifically talking about ETFs exposed to tech, these ETFs can prove more durable and offer greater upside.

Let’s talk about those investors who really do believe this is a bubble. An active tech ETF can adapt more quickly to issues than, say, than a passive fund can. An S&P 500 tech index, or even the S&P 500 itself, with market cap approaches, could take a bigger hit.

Consider a fund like the T. Rowe Price Technology ETF (TTEQ). TTEQ charges a 63 basis point (bps) fee to actively invest in global large cap tech firms. Managed by Dom Rizzo, mentioned above, the strategy can be “AI on” or “AI off” as required – giving exposure to the upside, but, if circumstances require, looking to other sources of innovation.

Adaptability like that can really help soothe uncertain clients about the state of the AI landscape. Overall, however, I would urge investors to look a bit longer term. Remember how much upside there still is in the AI landscape.

Finding Resilience Beyond the Hype

Shaw: TTEQ’s flexibility is valuable. An active manager who can rotate away from overheated AI names when needed makes sense. But there’s another approach. Avoid the concentration problem entirely by focusing on companies that never joined the spending race to begin with.

One example, Apple Inc. (AAPL), surged 35% between June and year-end 2025 while AI-focused peers like Meta Platforms Inc. (META) and Microsoft Corp. (MSFT) slid into the red, according to Bloomberg reporting.

Bill Stone, chief investment officer at Glenview Trust Company, told Bloomberg that Apple “avoided the AI arms race and the massive capex that accompanies it.” John Barr, portfolio manager at Needham Aggressive Growth Fund, noted it’s “remarkable how they have kept their heads and are in control of spending, when all of their peers have gone the other direction.” Apple now holds a $3.7 trillion market cap and the second-biggest weight in the S&P 500 Index.

For investors concerned about concentration risk in AI, equal-weight strategies offer another path. The Invesco S&P 500 Equal Weight Technology ETF (RSPT) automatically reduces exposure to the handful of hyperscalers dominating market-cap-weighted indexes, spreading risk across the broader tech sector.

The Range Nuclear Renaissance Index ETF (NUKZ) has emerged as a critical structural play. Tracking the VettaFi Nuclear Renaissance Index, the fund has returned 72.5% over the past year. NUKZ provides exposure across the nuclear value chain, including established utilities like Cameco Corp. (CCO) and GE Vernova Inc. (GEV) that provide the power AI data centers need, regardless of whether the software companies turn profits.

Beyond the power grid, advisors can look to sectors where AI is already being monetized to drive consumer margins. The Amplify Video Game Leaders ETF (GAMR) and the Amplify Online Retail ETF (IBUY) offer exposure to industries utilizing AI for real-time engagement and logistics. These ETFs allow investors to capture the result of the AI revolution through established revenue streams, rather than betting solely on the infrastructure race.

You’re right about the long-term upside. But investors can capture AI exposure via any number of ETFs, not just through an ETF focused strictly on AI.

For more news, information, and analysis visit the Thematic Investing Content Hub.

VettaFi LLC (“VettaFi”) is the index provider for UFO, ROBO, GAMR, NUKZ, BATT, ENFR, DRNZ, OGIG, and THNQ, for which it receives index licensing fees. However, those indexes are not issued, sponsored, endorsed or sold by VettaFi, and VettaFi has no obligation or liability in connection with the issuance, administration, marketing, or trading of those ETFs.

For more news, information, and strategy, visit ETF Trends.

Originally published on ETF Trends

A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.

Read more commentaries by VettaFi