‘magnificent seven’ stocks now account for 30 percent of the S&P 500’s total market cap
The sheer numbers are overwhelming. Analysts estimate that 15 to 25 percent of the S&P 500’s value can be attributed to expectations of AI delivering substantial financial benefits for companies, which translates into 800 to 1,300 index points.
However, if reality does not deliver on these expectations the market could correct to a level of around 3,900 to 4,400 points.
Accordingly, the fear among many analysts is that the AI investment boom – particularly in tech stocks – is showing the characteristics of a bubble. Let’s look at the evidence for and against bubble fears.
Why it could be a bubble
1. Valuations outpacing financial performance
- Companies with minimal AI revenue are seeing massive valuation premiums simply for mentioning AI.
- For example, the ‘Magnificent Seven’ tech firms (NVIDIA, Microsoft, Alphabet, Amazon, Meta, Tesla and Apple) now represent around 30 percent of the S&P 500’s total market cap, largely due to AI enthusiasm.
- Some stocks trade at forward P/E ratios above 50 to 70, which are levels seen during the dot-com bubble of the late 1990s.
2. Less short-term ROI
- While AI models are advancing rapidly, many companies haven’t yet found scalable, profitable applications.
- Conversely, a study by McKinsey claims that only 1 percent of companies are AI mature, that is, fully integrating AI into workflows and generating financial returns.
- Generative AI platforms like ChatGPT or Gemini generate huge traffic but still struggle to produce sustainable profits.
3. Expensive tech infrastructure
- Cloud providers and chip manufacturers are spending hundreds of billions on AI infrastructure.
- NVIDIA’s data center revenues, for example, have surged, but analysts warn of capacity overshoot if AI workloads plateau.
- When infrastructure investments outpace application maturity, bubbles can form, similar to the telecom fiber glut of 2000–2002.
4. Bandwagon behavior
- Companies in most sectors are claiming to be AI-driven to attract capital. On the other hand, concrete data on the financial impacts of their AI engagements are few and thin.
5. Similar bubbles in the past
- Past bubbles:
- Railway bubble (1840s)
- Dot-com bubble (1990s)
- MBS/CDO bubble (2007)
- In the first two cases, technological revolutions were real – but expectations overshot reality.
6. A lot of leverage
- AI-related debt has accounted for $141 bn in corporate credit issuance in 2025 so far, eclipsing full-year 2024 gross supply of $127 bn.
The countervailing view is that we’re not in an AI bubble, but that we are in the midst of an ‘AI revolution’, whereby the technology should be considered a fundamental game changer akin to those of electricity and the World Wide Web.
1. Real AI impacts
- AI is already driving productivity and revenue across industries – not just in tech:
- Microsoft Copilot and Google Gemini are increasing employee efficiency.
- Amazon, JPMorgan, and Siemens use AI for supply chain optimization, cutting logistics costs by billions.
- Unlike crypto or past hypes, AI’s value creation appears tangible, measurable and scalable.
2. Companies expected to adopt AI quickly
- More than 80 percent of Fortune 500 companies have dedicated AI initiatives in production or scaling phases.
- Spending is tied to areas able to deliverable financial results:
- Process automation
- Customer service (AI agents, chatbots)
- Predictive analytics
- Risk management.
3. Lasting, necessary tech infrastructure
- The massive investments in GPUs, data centers, and AI chips are laying the foundation for the next decade of computing.
- Unlike the dot-com era, today’s build-out has:
- Immediate commercial use cases.
- Real capacity utilization (eg OpenAI, Anthropic, Meta or Tesla).
- These assets (chips, data centers) retain long-term economic value, not speculative paper wealth.
4. AI driven revenue growth and margins
- Given that a large part of the market overvaluation is for the big tech firms, a bubble may not be broad based. Tech companies revenues have truly exploded. NVIDIA’s data center revenue alone grew up 279 percent year-over-year (2024).
- Microsoft, Amazon and Alphabet report AI-related revenue segments expanding faster than any other. NVIDIA, Microsoft and Alphabet together generate more than $200 bn in annual profits. Generative AI services (ChatGPT, Claude, Gemini, Midjourney) already account for billions in average rate of return terms.
5. Governmental support
- Governments view AI as strategic infrastructure:
- Just think of the EU AI Act, US CHIPS and Science Act, China’s AI Strategy or the UK’s AI Safety Summit.
- National security, competitiveness, and productivity all depend on AI.
- This policy support ensures sustained long-term investment and adoption.
6. AI use & deployment faster than past tech
- Internet took around 10 years to reach mass adoption; AI tools hit 100 million users in under two months.
- Network effects are accelerating innovation cycles.
- The faster diffusion curve suggests a true technological inflection point, not a transient mania.
- But while popular diffusion of popular AI such as ChatGPT is rapid, corporate deployment is another story.
In any case, at least three salient facts stand out:
- Valuations of too many companies are off the charts and typically will correct. Projects about AI growth are largely for the next five years, yet many AI-related private equity funds project earnings 20, 30 and even 50 years into the future, thus revealing highly speculative investment behavior.
- Companies are slow to deploy AI. Supposedly only percent are actually successfully deploying AI while 98 percent feel a sense of urgency to adopt AI, according to Cisco’s AI Readiness Index.
- Transparency about actual AI deployment and its success and failure in companies is severely lacking. Assessments of AI deployment in companies are based mainly on PR and speculation.
Thus, we may be looking at small bubble bath bubbles popping instead of a big explosion.
IR executives should establish a reporting framework on their companies which is structured, heavy on numbers:
- Deployment plans
- Timelines
- Costs
- Expected financial impacts, covering revenues, margins and cost of capital, among others.
Dr William Cox has consulted more than 50 blue chip companies and institutional investors over the course of decades on the financial impacts of corporate process, ESG and AI. He received his PhD from the London School of Economics and other degrees from Oxford, Boston and the Harvard Kennedy School.
