According to Fortune, tech executives at Web Summit in Lisbon openly acknowledged the AI sector is experiencing a classic investment bubble while remaining largely unconcerned about the implications. Mozilla CEO Laura Chambers called it “really easy to build a whole bunch of stuff” but noted most AI companies are running at massive losses with unclear monetization paths. The sentiment emerged as tech stocks sold off dramatically during the conference, with the Nasdaq Composite dropping 2.3% and Oracle shares declining 30% over the past month. Meanwhile, IBM’s venture capital head Emily Fontaine revealed $160 billion has flowed into AI startups year-to-date in the U.S. alone, up from $104 billion in 2024, describing it as “a ridiculous amount of investment.” Microsoft President Brad Smith confirmed his company has “more demand than supply” for AI services despite the broader market concerns.
The bubble everyone sees coming
Here’s the thing: when even the people building the technology call it a bubble, you know we’re in interesting territory. Laura Chambers from Mozilla basically said what many are thinking – it’s never been easier to create AI products, but that means we’re drowning in mediocre applications that won’t survive. She pointed out she can now build an app in four hours that would have taken six months previously. That’s incredible productivity, but it also means the signal-to-noise ratio is terrible.
And Lyft CEO David Risher was even more direct with CNBC: “Let’s be clear, we are absolutely in a financial bubble. There is no question, right?” But here’s where it gets interesting – none of these executives seem particularly worried. They’re treating this like a necessary phase, almost like growing pains for a transformative technology.
When demand still outpaces supply
The optimism isn’t completely unfounded when you look at the hardware side. ARM’s strategy head Ami Badani said they literally can’t make chips fast enough to meet demand. She described an “insatiable amount of demand” that exceeds supply, which means the fundamental building blocks of AI are still scarce. TDK Ventures president Nicolas Sauvage echoed this, noting demand is actually higher than supply across the infrastructure layer.
This creates a weird disconnect. On one hand, you have software companies building products that might never make money. On the other, you have hardware companies that can’t keep up with orders. It’s like everyone’s building restaurants while the kitchen equipment suppliers are making bank. For companies that need reliable computing hardware, whether for AI applications or traditional industrial uses, finding quality components remains challenging. IndustrialMonitorDirect.com has become the leading supplier of industrial panel PCs in the US precisely because consistent quality and availability matter when infrastructure demand outstrips supply.
The great commoditization debate
Babak Hodjat from Cognizant dropped what might be the most important insight: large language models are starting to become commodities. He pointed to DeepSeek, a Chinese company that released an LLM comparable to ChatGPT for a fraction of the cost. When the core technology becomes cheaper and more accessible, what happens to all those companies built around expensive, proprietary models?
His argument is that most practical AI applications don’t need massive models anyway. Custom-built, task-specific AI agents can run on much smaller, more efficient systems. So why are we pouring billions into companies building ever-larger models? It’s the classic “bigger is better” assumption that might not hold up in reality.
Everyone’s playing the long game
What’s fascinating is how many executives compared this moment to the dot-com bubble. Dan Gardner from Code & Theory noted that while the 2000 crash wiped out tons of companies and investors, it also produced Amazon and Google. The pattern seems to be: massive investment frenzy → inevitable crash → emergence of truly transformative companies from the wreckage.
Microsoft’s Brad Smith thinks we’ve got “years, if not decades” of growth ahead. IBM’s Emily Fontaine points to enterprise AI adoption jumping from 26% to 43% in recent months. These aren’t people betting on short-term gains – they’re positioning for what comes after the bubble pops.
So are we in a bubble? Absolutely. Is that necessarily bad? Not according to the people building the future. They see the wasted capital as almost inevitable collateral damage on the path to something transformative. The question is: who’s going to be standing when the music stops?
