Photo by Steve Johnson on Unsplash

The rapid rise of artificial intelligence has naturally sparked questions among investors. Some worry that the extraordinary spending and explosive revenue growth are part of a self-reinforcing bubble, especially given diagrams showing financial connections between Microsoft, OpenAI, Nvidia, Anthropic, Amazon, and Oracle. While those visuals are attention-grabbing, the reality is far more grounded. When you combine audited financial results with the massive industrial buildout currently underway, it’s clear that AI growth is built on real revenues, genuine cash flow, and substantial third-party economic activity. Moreover, this is only the first phase: the systems being constructed today lay the groundwork for the eventual pursuit of Artificial General Intelligence (AGI).

The AI Bubble Argument

Some analysts argue that today’s AI boom shows classic signs of a bubble, pointing to valuations that far outpace fundamentals, weak monetization, and extreme concentration risk. Many AI startups are priced at 20–50x revenue, with outliers well above 100x, and some public AI-themed companies trade at hundreds of times earnings. Corporate adoption data also raises concerns: studies suggest that the vast majority of organizations investing in generative AI are not yet seeing measurable returns, even as global AI startups raised more than $70 billion in a single quarter and large tech firms commit hundreds of billions of dollars to data centers, GPUs, and energy infrastructure. Bears argue this imbalance — massive capital in, limited revenue out — mirrors early-stage bubbles where investors pay for potential rather than proven profitability.

“95% of organizations get zero return on GenAI pilot.” - MIT Report

“Median revenue multiple for AI companies stood at 29.7x.” - Adventis Advisors

“80% of U.S. stock gains this year came from AI companies.” - Brookings, Is there an AI bubble?

Skeptics also flag structural risks beneath the surface. A small cluster of mega-cap AI leaders drives a disproportionate share of market gains, leaving indexes vulnerable if sentiment shifts. AI models require extraordinary amounts of compute, water, and power, raising the possibility of overbuild or stranded infrastructure if expectations come down. Some critics even worry about “circular demand,” where startups raise money primarily to buy more chips, fueling the very revenue growth used to justify high valuations. Taken together, these factors — stretched multiples, limited ROI, concentration, and massive capex — form the core of the argument that parts of AI investing may be ahead of themselves, even if the long-term technological opportunity remains real.

Speculative Investment Behavior

November’s market activity largely reinforced the trends I highlighted last month. November has seen a notable pullback in the markets, with the Nasdaq down roughly 3% so far. Even standout earnings have not been immune to selling pressure: Palantir and Nvidia both reported impressive results, but their shares sold off afterward, reflecting the market’s heightened sensitivity to expectations and valuations. This pattern highlights the narrow, sentiment-driven nature of recent gains, where even strong fundamentals can be overshadowed by investor psychology. Volatility has also been evident in cryptocurrencies, with Bitcoin down about 20% over the month, underscoring the rapid swings that can occur in highly speculative assets. Overall, while earnings and fundamentals remain strong for many leading companies, these recent moves reinforce the importance of maintaining a disciplined, diversified approach to investing.

Real AI Investment

Not all AI revenue is hypothetical or internally recycled. Companies across the AI supply chain are reporting cash-generating sales tied to real customer usage. Nvidia, for example, generated $51.2 billion in data center revenue in a single quarter, fueled by global demand for GPUs, networking systems, and full AI computer racks purchased by enterprises, governments, and cloud providers. These are tangible goods acquired with cash, not internal accounting loops.

Related podcast: Mark Mills: AI Most Energy-Hungry Tech Ever Invented – Data Centers, the Grid, and Black Swans

Cloud providers show a similar story. Microsoft’s fiscal 2025 operating cash flow totaled $136.2 billion. Google Cloud added billions in quarterly revenue as real businesses pay for computing power to run new models and applications. All these figures are audited and reflect actual commercial transactions.

Show Me the Money

Talk is cheap, so here are the numbers from the latest third quarter earnings season:

• Nvidia – Total revenue of $57B, up 62% from last year; data-center revenue $51.2 in a single quarter.
• Advanced Micro Devices – $9.2B in revenue last quarter, up 36% from a year ago, and free cash flow of $1.5B.
• Google – Consolidated revenue $102.3B and $155 billion in backlog; cloud revenue $15.2B with free cash flow of $24.5B.
• Microsoft – Cloud revenue of $49.1B, up 26%, with cash from operations of $136.2B.
• Amazon / AWS – AWS revenue of $33.0B (up 20%).
• Cisco reported $2B for AI-related revenue in 2025 and expects $3B in 2026.
• GE Vernova revenues of $9.97B in Q3, up 12% over last year and year-to-date free cash flow of nearly $2B.
• Eaton revenues of $7B last quarter, up 10% with free cash flow of $1.2 billion - third quarter records.

Source

AI Spurs One of the Largest Buildouts in Modern History

AI cannot scale without significant physical infrastructure. The current investment wave focuses on hardware, factories, energy, and data centers, which require billions in equipment, construction, land, and long-term power agreements. Tawain Semiconductor Manufacturing, for instance, plans $40–$42 billion in capital spending to expand semiconductor production, while ASML recorded €7.5 billion in quarterly sales from advanced chipmaking machines used to produce AI processors. These are real, global transactions involving independent suppliers and labor, far beyond any internal investor network.

See also: Cracks in the AI Trade: China Heats Up the Competition, Says Peter Boockvar

Hyperscale cloud providers such as Amazon, Google, Microsoft, and Oracle are collectively investing hundreds of billions to build data centers, driving spending on steel, concrete, fiber networks, cooling systems, and thousands of workers. By 2030, Oracle alone is estimated to be committing $300 billion toward AI-focused data center expansion according to Bank of America’s analyst, Brad Sills. There is concern about a software company committing to such significant, fixed investments and the potential impact this could have on its profit margins. Sills also highlighted Oracle’s smart way of financing and maintaining grounded economics for its contracts adding “the company structures its AI compute agreements as non-cancelable, take-or-pay contracts that guarantee stable, predictable cash flows and cost visibility—unlike rival hyperscalers building speculative capacity” in an article on Yahoo on October 28. This industrial activity represents genuine economic growth, not a self-contained circle of investments.

Data Center Construction Outpaces Traditional Commercial Projects

Data center construction is currently growing faster than conventional commercial construction, reflecting AI and cloud demand. U.S. spending on data center construction surged 92.8% through September 2025, reaching $32.9B according to Westside Construction Group on November 6^th in their article, “Data Center Construction Boom Reaches $32.9B”. They note the average data center project costs $193 million, but they are seeing billion-dollar facilities becoming more common. The average in monthly starts is $3.5 or more billion for three consecutive months as of their article’s date on November 6^th.

Operating Data Centers

Source: U.S. Department of Energy

Planned Data Centers

Source: U.S. Department of Energy

While office and retail construction have slowed, data centers account for most of the growth in non-residential construction, alongside manufacturing. Their scale, complexity, and need for specialized contractors, institutional financing, and accelerated timelines distinguish them from traditional commercial projects.

This surge reflects the predictability of demand from hyperscalers and the strategic prioritization of digital infrastructure over conventional commercial expansion. While still a small share of total commercial construction, the rapid growth of data centers highlights a broader shift toward building the infrastructure required to support AI and cloud computing. Investors should also note potential constraints, including grid capacity, labor competition, and regulatory challenges that could influence project timelines and costs.

Source

Why Valuations Can Still Make Sense

Despite lofty AI valuations, even traditional value investors are taking notice. Warren Buffet’s Berkshire Hathaway’s recent $4.3 billion investment in Alphabet (Google) in the third quarter and Cathie Wood’s return to Nvidia illustrates that real revenues, strong cash flow, and strategic infrastructure remain compelling. Alphabet’s massive AI data centers, scalable cloud operations, and consistent monetization provide a foundation that goes beyond hype. For investors, this shows that participation in companies with proven earnings and infrastructure can offer growth exposure while retaining resilience in a rapidly evolving AI economy.

Must read: Beyond AI: How Blockchain and DAOs Will Reshape the Future of Work

Phase 1 Is Just the Beginning

The current industrial expansion is not aimed at creating Artificial General Intelligence (AGI) today but at laying the foundation for it. Today’s AI systems are powerful but narrow, excelling at specific tasks such as coding, summarizing documents, or generating images. AGI, by contrast, will require systems capable of reasoning, learning, and adapting across a broad range of tasks in a human-like way.

Reaching AGI will demand enormous computing capacity, energy, advanced chip architectures, persistent memory systems, and reliable, interpretable models. The scale of today’s investment reflects preparation for that second phase, not overbuilding. Data centers, semiconductor facilities, and energy infrastructure are the groundwork for a far more ambitious technological leap than the internet or smartphones ever required.

What Will AI Look Like in 10 Years?

If current progress continues, AI use in 10 years will likely feel as normal and embedded as smartphones are today—powerful, ever-present, and woven into nearly every workflow, but operating mostly in the background.

For individuals, AI will likely function as a continuous personal assistant that understands long-term goals, manages schedules, drafts communications, handles routine digital tasks, and coordinates across devices without being prompted. Search may shift from typing queries to having ongoing conversations where the system already knows your preferences, context, and history. Much of everyday software—email, documents, photo tools, operating systems—will have intelligent copilots built in, not as add-ons but as core functionality.

For businesses, AI will probably run large parts of operations: automated compliance checks, real-time financial analysis, forecasting, customer service, and dynamic resource management. Knowledge workers will spend far less time on preparation and far more on decision-making. Entire departments may operate with “AI first”—where humans guide strategy while AI handles the data, analysis, and execution.

On the frontier, the early foundations of AGI-like systems could emerge. These would not replace human judgment, but they may reason across domains, monitor complex systems, and collaborate on multi-step projects. Industries like medicine, energy, manufacturing, and finance may see the biggest shifts, with AI helping diagnose disease, optimize infrastructure, and design new materials or investment strategies.

Overall, AI in a decade is likely to be more natural, more context-aware, and deeply integrated into daily life—less a tool you “use” and more a background capability that quietly amplifies what individuals and organizations can accomplish.

Conclusion

Claims that AI growth is merely a bubble ignore the broader picture. Revenues are real, cash flow is strong, and global demand is deep and diverse. The physical buildout under way is one of the largest in decades, spanning chips, data centers, networking, and power. Today’s expansion represents Phase 1, setting the stage for Phase 2: the pursuit of AGI. Understanding this context helps investors appreciate why leading global companies are committing so heavily to AI infrastructure—and why the systems being built today are just the beginning of a far larger technological cycle.

While many companies within high-growth thematic areas such as artificial intelligence are generating substantial revenues and real earnings, it is important to remember that not all opportunities are created equal. Speculative investing—particularly in firms that have yet to achieve net income and are largely selling visions of future potential—carries significant risk. Investors should conduct thorough due diligence, understanding both the fundamentals and the valuations of the businesses they consider. Growth themes can be compelling, but success comes from investing based on solid financials and realistic prospects, not chasing hype. Don’t just chase the hottest name you hear from a vlogger on YouTube. Always approach emerging sectors with discipline, focusing on investments rather than speculation.

More By This Author:

From Fiscal Band-Aids To AI Shakeout: Why Markets Remain On Edge
How AI, Rate Cuts, And Commodities Are Shaping The Market
The High Cost Of Our Digital Lives: Unveiling The “Dark Cloud”