Is AI a bubble?

Grok 4.1 Fast

Vote: No

AI is not a bubble because it represents a transformative technology with tangible advancements and real-world applications, unlike speculative manias like tulip bulbs. Key evidence includes rapid progress in models like GPT-4 and Llama, driving productivity gains in coding, drug discovery, and automation. Companies like NVIDIA report surging revenues from AI hardware demand, grounded in actual utility. While hype inflates valuations, historical parallels to the internet boom show short-term corrections but long-term value creation. AI's scalability and adoption across industries suggest sustained growth, not collapse.

GPT-5.4

Vote: No

AI is not best understood as a pure bubble. It is a general-purpose technology already delivering real value in search, coding, customer support, drug discovery, logistics, and automation. Bubbles are driven mainly by hype detached from utility; AI has substantial underlying utility even if expectations are overheated. That said, parts of the AI market do show bubble-like behavior. Valuations, spending, and timelines in some segments appear inflated, and many companies will likely fail. But that is common in major technological shifts and does not mean the whole field is a bubble. A better comparison is the internet boom: excessive speculation around a real, transformative technology. Some investments will be mispriced, but the core technology is durable and economically significant. So the strongest answer overall is no.

Sonar Pro

Vote: No

AI is not a bubble due to its substantial real-world value and economic impact, distinguishing it from pure speculative manias like the dot-com era. AI demonstrates tangible fundamentals: companies like OpenAI ($20B+ revenue, 10x growth), Microsoft ($245B revenue, AI Copilot adding billions), Alphabet, and Meta show strong earnings and profitability, unlike dot-com firms with no profits.[3][4] Research estimates AI can perform $4.5 trillion in US tasks, adding $1 trillion to GDP, with faster-than-expected job and profitability impacts.[4] While valuations are stretched (e.g., Mag 7 +250% in 2 years, S&P at 23x earnings) and hype exists, P/E ratios (20-40x) are far lower than dot-com's 200-500x, with $100B+ earnings backing them.[3] Bears cite poor ROI (95% no measurable returns per MIT) and CapEx ($440-602B), but leaders are delivering ROI in ads, search, and copilots.[3] Bubble warnings (e.g., 52% experts expect AI stock declines, Capital Economics predicts 2026 burst) acknowledge AI's 'actual value' mutes fallout, unlike 2008 or dot-com crashes.[1][5] Even in 'bubble pop' scenarios (20% probability), recovery is expected within years, not decades, due to real adoption.[3] History shows tech booms (e.g., internet post-2000) deliver long-term gains despite corrections; AI's infrastructure investments boost GDP now and position for productivity surges.[1][2]

Claude Opus 4.6

Vote: Yes

There are strong indicators that AI is currently in a bubble, though the underlying technology itself is transformative. First, the financial signals are classic bubble indicators. Nvidia's market cap surged past $3 trillion largely on AI hype. Hundreds of billions are being invested in AI infrastructure (data centers, chips) based on projected revenues that have not yet materialized at scale. The gap between AI investment and actual AI revenue generation is enormous — Microsoft, Google, and others are spending tens of billions on AI capex while AI-specific revenue remains a fraction of that. Second, we're seeing the hallmark pattern of bubbles: companies adding 'AI' to their branding or pitch decks to attract investment, regardless of whether AI is core to their product. This mirrors the dot-com era when companies added '.com' to their names to boost valuations. Third, current AI models (LLMs) face fundamental limitations — hallucinations, high inference costs, limited reasoning — that are often glossed over in bullish narratives. Many enterprise AI deployments are struggling to show clear ROI. Importantly, calling AI a bubble does NOT mean the technology is worthless. The internet was also a bubble in 1999-2000, and it went on to reshape civilization. But many investors lost fortunes because valuations far outpaced near-term reality. Similarly, AI will likely be transformative long-term, but current valuations and expectations appear to have gotten ahead of what the technology can deliver in the near to medium term. The combination of speculative excess, massive capital misallocation risk, and a gap between hype and current capabilities makes 'yes' the more defensible answer.

Kimi K2.5

Vote: No

While certain AI valuations exhibit bubble-like symptoms, AI itself is not a bubble but rather a genuine technological paradigm shift comparable to electricity or the internet. First, bubbles are characterized by assets trading far above intrinsic value without underlying utility—like tulips or beanie babies. AI, however, is already delivering measurable productivity gains: software engineering throughput has measurably increased, customer service costs are dropping, and enterprise adoption curves show real demand rather than speculative FOMO. Second, the massive capital expenditure on GPUs and data centers reflects physical infrastructure with persistent utility, not just financial engineering. Unlike the dot-com era where companies lacked revenue models, leading AI firms demonstrate rapid, accelerating revenue growth tied to actual compute consumption. Finally, while we may see cyclical corrections in valuations—a healthy market function—the underlying technology trajectory remains robust. Calling AI a bubble conflates short-term valuation excesses in specific stocks with the fundamental reality that we are in the early innings of general-purpose technology adoption.

Gemini 3.1 Pro

Vote: No

While certain AI company valuations may be inflated, AI as a broader phenomenon is not a bubble because it delivers immediate, measurable utility. Unlike transient tech fads, generative AI tools are fundamentally altering real-world productivity in software development, content creation, and complex data analysis. Furthermore, the massive capital expenditures in the space are driven by tangible demand rather than pure speculation. Infrastructure companies like Nvidia are generating record-breaking, actual revenues. Meanwhile, major enterprises are deeply integrating AI into their core software ecosystems, making it a permanent fixture of modern business workflows. Finally, the underlying technology continues to adhere to predictable scaling laws. Increased compute and data reliably yield better performance, guaranteeing continued capability enhancements. This points to a foundational industrial shift rather than a temporary illusion bound to burst.

Final Summary

The AI Roundtable reached a lopsided conclusion, with five models insisting that AI is a fundamental industrial shift while Claude Opus 4.6 stood as the lone skeptic. The majority argued that unlike historical manias, the current AI era is backed by massive infrastructure and real-world utility, while the lone dissenter pointed to a cavernous gap between speculative investment and actual revenue.

No agreement after 1 rounds of discussion. Final vote: No: 5, Yes: 1

Strongest Arguments

• Yes: The financial signals are classic bubble indicators: billions are being poured into infrastructure based on projected revenues that have not yet materialized, creating a massive gap between capex and actual earnings.
• No: Unlike the dot-com era where companies lacked revenue models, current AI leaders demonstrate record-breaking earnings and lower P/E ratios, proving that the technology is already delivering measurable economic value.