AI Bubble: Speculation, Sustainability, and What Comes Next

Emily Carter

Hello and welcome back to Global Equity Research, the podcast where we break down the latest trends in finance and macroeconomics, all brought to you by Softgate Capital Research. I’m Emily Carter, here as always with David Mitchell.

David Mitchell

Hey folks, David here. Before we get started, I want to thank everyone out there for sticking with us all year—your questions, your feedback, your support, all the likes and shares, it means a lot. Seriously, we wouldn’t be here without you.

Emily Carter

Yeah, a huge thank you, and just a reminder—Softgate Capital Research is the research arm of Softgate Capital, and our mission is to provide institutional-grade research for everyone. We want insightful, actionable analysis to be accessible, not hidden behind those big price tags.

David Mitchell

And heads up—this is our last episode for 2025. I know, it’s wild to say that! But don’t worry, we are absolutely coming back next year with even more analysis and deep dives. So you can expect us back in 2026, recharged and ready.

Emily Carter

Today is a special one, too—a topic that’s had markets buzzing, investors nervous, and think pieces coming out of our ears. We’re digging deep into the debate over whether there’s an AI bubble. And as always, you can find the full special report for free in the description or at research.softgatecapital.com. No paywall, just good research. Ready, David?

David Mitchell

Ready as I’ll ever be. So, zoom out for a second and just think about what’s going on: companies like Nvidia, Microsoft, OpenAI—they’ve just been on an absolute tear. Market caps hitting the trillions, record funding flowing to any startup that can spell “AI,” and headlines promising that every problem on Earth is solvable with enough algorithms. Feels familiar? Because, you know, a lot of people are starting to say this all looks a bit, well, bubbly.

Emily Carter

Right, I mean the level of hype is just enormous. But is it hype, or is it that big technological shift we only see maybe once in a generation? There’s so much emotion in this debate—and plenty of strong data, too. We’re gonna try to cut through the noise today.

Emily Carter

So, before we get carried away, let’s actually define what we mean when we talk about an “AI bubble.” A bubble in finance is pretty simple on the surface: asset prices go way up, often because of speculation, wild expectations, and usually some classic FOMO, and then—pop—it all comes crashing down, hard.

David Mitchell

Exactly. And in the context of AI, that means we’re talking about valuations for AI-related companies and assets skyrocketing way past their true, or you could say intrinsic, value. A lot of it’s driven by excitement—or exaggerated expectations—rather than delivered business results.

Emily Carter

You’ll recognize some of the classic signs: huge price jumps, crazy venture funding rounds, IPOs everywhere, and media headlines screaming ‘AI Revolution’ like it’s the dawn of a new era. Oh, and let’s not forget the regulatory frameworks—often lagging way behind what’s happening in the markets. Sound familiar?

David Mitchell

All too familiar. That sort of combination—hype, runaway values, lagging oversight—is what keeps both investors and regulators up at night.

David Mitchell

Let’s talk about why so many experts are waving the bubble flag right now. First, just—look at the numbers. Nvidia’s market cap is up over 1300% since late 2022. By the end of 2025, it’s the world’s most valuable company. Valuations for OpenAI, Anthropic, even unicorns like Databricks—it’s all eye-popping. Venture capital in AI? $202 billion this year, half of all global VC funding. That’s not a typo—half.

Emily Carter

And it’s not just the size, it’s the concentration. The “Magnificent Seven”—that’s Nvidia, Microsoft, Amazon, Google, Meta, Apple, Tesla—they’re responsible for more than a third of all S&P 500 performance in 2025. Seventy-five percent of market returns are coming from just 41 AI stocks. If that’s not concentration risk, I don’t know what is.

David Mitchell

Valuation multiples are nuts. AI startups routinely raise at revenue multiples of 25x, 35x—even 45x. Compare that to traditional SaaS, which is more like 6x to 8x. On some metrics, we’re actually seeing extremes that even the dot-com bubble didn’t touch, especially when you look at market caps or valuation per employee—hundreds of millions per head isn’t unusual now.

Emily Carter

And let’s talk debt and spending. The hyperscalers—Microsoft, Amazon, Google, Meta, Oracle—are expected to spend a trillion dollars on AI by 2026. OpenAI alone is planning to drop $1.4 trillion in just three years. Data center costs, resource strain—huge. And a lot of this is funded through record bond issuance. More leverage, more risk if those investments don’t pan out.

David Mitchell

Yet, despite all the spending, profitability is lagging for many companies, especially in generative AI. OpenAI expects to lose $9 billion on $12 billion in revenue this year—cumulative losses in the tens of billions projected. Bain & Co. estimates we’ll need $2 trillion in AI revenue annually by 2030 just to justify the current capex, and we may fall $800 billion short of that.

Emily Carter

And you have the behavioral side—FOMO’s everywhere, institutions and retail are piling in, sometimes with very little understanding. In a survey last October, 54% of fund managers said AI stocks were deep in bubble territory. It’s not just the numbers, it’s also the mood. It feels—frothy.

Emily Carter

I want to connect these dots to history. The dot-com bubble, everyone remembers—or maybe not, but we’ve all read about it—mid ‘90s to 2000, internet startups booming on pure optimism, most of them unprofitable, endless hype. The Nasdaq forward P/E hit 60x. Only 14% of IPOs were profitable at the top. Eventually, it all collapsed.

David Mitchell

The parallels with AI today are spooky, honestly. Huge capital flows into companies, even those with sketchy business models. Unicorns multiplying overnight, IPO windows flying open. We just did a whole episode on Nvidia’s rise from gaming GPUs to the world’s AI backbone—if you missed that, rewind, because it’s a case study on how rapidly markets can reprice a company on the back of a technological wave.

Emily Carter

But there are differences too. While dot-com P/Es were crazier than today’s tech leaders—Nasdaq-100 forward P/E was about 60x, now it’s 26x—other things, like the sheer size and concentration, are bigger this time. Oh, and data centers and infrastructure actually exist now. There’s physical investment behind the numbers, not just paper IPOs.

David Mitchell

One quick stat here just for context: there were 47 new AI unicorns in 2024, and AI startups are commanding valuations per employee that are truly wild. But, uh, like you said, Emily, at least some of these companies are actually generating billions in sales. That's a key difference.

Emily Carter

On the flip side, let’s not ignore what the skeptics say. A lot of smart people argue this isn’t really a bubble—at least, not in the old sense. AI is seen as a foundational technology, like electricity or the internet, and by late 2024, 78% of companies worldwide are already using it. Enterprise adoption is real, with massive productivity and cost gains starting to come through, especially in big established firms like Walmart and Goldman Sachs.

David Mitchell

Exactly. Unlike the dot-com era, the leading AI companies—Nvidia, Apple, Microsoft, Google—are genuinely profitable. Nvidia, for example, has a net margin over 53% and is doubling revenue year-on-year, which, I mean, that’s just—it's astonishing. The hyperscaler cloud divisions, like Google Cloud and Azure, are reporting big jumps in AI services and profits too.

Emily Carter

And the investment isn’t all vaporware this time. There’s enormous spending on things you can actually touch: data centers, GPUs, physical infrastructure. Even if some projects fail, the legacy assets are likely to power future productivity gains and new waves of technology. It’s not money just tossed at a shiny website hoping for eyeballs.

David Mitchell

Plus, let’s not forget the guardrails. There’s the EU AI Act, state-level regulation in the U.S., and a lot of public-private partnerships for transparency, safety, and actual governance. Central banks, sovereign funds—they’re all way more active in monitoring systemic tech risk than they ever were in 2000.

Emily Carter

And finally, some theorists argue that high valuations aren't always irrational—there’s this idea of a “speculative growth framework,” where today’s crazy-looking prices might actually make sense if they accelerate capital formation for transformative tech. It’s fragile, but maybe not just hype.

Emily Carter

Let’s play out the scenario: what happens if this AI bubble—if it is a bubble—actually bursts? There are early warning signs we know to look for: valuations shooting up faster than real earnings, institutional investors quietly cashing out while retail’s still piling in, leverage getting high, liquidity starting to tighten as banks or brokers raise collateral demands.

David Mitchell

Companies missing earnings can be a big one; you get a bad report from a leader, like Nvidia, and suddenly you see, I don’t know, $600 billion in market cap wiped out in a single day—like what happened after DeepSeek’s launch. Another trigger could be a hiccup in the massive infrastructure fueling AI, say a big data center outage, or companies struggling to pay back all this new debt.

Emily Carter

And then there’s institutional rebalancing. When big funds or pensions start quietly moving away from the crowded corners of the market, that can make things unravel faster than people expect. The bottom line is, a lot of these warning signs are what you’d expect to see before a big correction—maybe soft, maybe hard, and maybe worse.

David Mitchell

And it’s not just about stock prices. If we do get a hard landing—let’s hope not, but if we do—the fallout spreads way beyond the financial markets. There’s the economic risk: a crash in AI could easily spark broader market volatility, reduced investment, and potentially recession, especially since AI investments have been juicing GDP growth for a while now.

Emily Carter

There are also industry-specific effects. Cloud providers and chipmakers could end up with massive overcapacity—data centers built on growth projections suddenly go underutilized, hardware demand dries up, and startups struggle to get new funding. I mean, we saw something similar with the rare earth sector last year, and if you missed our episode on that, by the way, go check it out. But this could be much bigger.

David Mitchell

There’s a labor market angle, too—AI can displace jobs in routine, automation-prone sectors, but it also creates new roles. Still, if funding dries up across the startup ecosystem, a lot of early-career folks, especially entry-level, could get hit hard. Many of those “AI trainer” or “prompt engineer” jobs are pretty new, and could be first to go if there’s a deep freeze.

Emily Carter

Plus, there’s always the risk of an “innovation winter.” If things get too ugly, capital flees, R&D slows way down, and we risk another period of stagnation, just when everyone’s getting used to rapid AI progress. Public trust also matters—if sentiment turns badly, regulation will probably come in even heavier, and it could take years to regain momentum.

Emily Carter

So, David, what’s your call? Are we looking at a bubble that’s about to pop—or is this just what the early days of a foundational technology transition look like? The “speculative growth framework” theory suggests bubbles aren’t always signs of irrational excess. Sometimes, high prices are the fuel that lets these epic technology shifts get off the ground.

David Mitchell

Yeah, and I’d say the answer isn’t very satisfying—it’s kind of both, right? There's definitely froth, some classic bubble signs, and pockets of dangerous speculation. But there are also real business models, real profits, and real tech breakthroughs happening. So maybe we’re living through a transition period, where speculation and substance are kind of tangled together and it could fall either way.

Emily Carter

I like that framing—multiple equilibria, to steal the economist jargon. Markets can stay irrational longer than we think, but sometimes that sets the stage for genuine long-term value. The hard part is staying clear-eyed about which is which, especially as investors.

David Mitchell

So, to wrap up: yes, we see classic bubble markers—sky-high valuations, concentrated returns, speculative frenzy. But this is definitely not a replay of dot-com, because the tech’s more integrated, the profits are more real for the leaders, and the infrastructure actually matters.

Emily Carter

A bubble pop could have sweeping implications, from market corrections and debt crises to innovation slowdowns and job market churn. But broad adoption and physical investment might be what cushions the blow and sets the stage for a more sustainable, if slower, next wave. Ultimately, it’s about distinguishing hype from meaningful advances, and regulators, policymakers, and businesses all have a role here.

David Mitchell

Couldn’t have said it better. As always, thanks so much for tuning in all year. If you found this helpful, please consider liking, subscribing, sharing—we’ll be back in 2026 with even more analysis and, I dunno, maybe a little less froth? Hopeful thinking, right?

Emily Carter

Yeah, here’s to smart investing and great profits in 2026. Thanks again for all your support, and don’t forget to grab the full report—free—from the description or research.softgatecapital.com. David, always a pleasure to dig into this stuff with you.

David Mitchell

Likewise, Emily. Take care, everyone. See you next year!

Emily Carter

Bye everyone. Happy holidays, and stay curious!