The AI Gold Rush: Who's Actually Making Money (And Why It's Not Who You Think)

You might think the AI gold rush is minting fortunes for companies like OpenAI and Anthropic, but the numbers tell a different story. Tech giants are expected to invest close to $700 billion in a single year.

The AI startups getting all the headlines are spending money faster than they are earning it. The real winners? The companies that sell the infrastructure. In fact, thousands of newcomers have driven the AI data centre gold rush, creating massive profits for cloud providers and chip makers. This pattern mirrors every tech boom in history. Understanding it could reshape your investment strategy.

The Gold Rush Analogy: Why History Repeats in Tech

The Original California Gold Rush Lesson

Gold fever gripped America when James W. Marshall found gold at Sutter's Mill on January 24, 1848. Over the next seven years, approximately 300,000 people flooded into California, each convinced they'd strike it rich. Most didn't.

Recent scholarship confirms that merchants made far more money than miners during the gold rush. Samuel Brannan became the first millionaire of the rush not by panning for gold but by buying mining supplies and reselling them to desperate prospectors. He earned about $36,000 in a short period, equivalent to roughly $1.5 million today.

Levi Strauss was another winner who opened a dry goods company in San Francisco in 1853. Tailor Jacob Davis proposed partnering to patent riveted work pants in 1872, and Strauss agreed. On May 20, 1873, they received U.S. Patent No. 139,121 for what would become blue jeans. Strauss never chased gold. His estate was worth nearly €5.73 million at the time of his death in 1902, about €166.99 million in current value.

Who Made Money Then vs Now

The pattern holds over centuries. Merchants who sold picks, shovels, tents and durable clothing built lasting fortunes while most miners went home broke. This "picks and shovels" strategy works because it reduces exposure to extraction risk.  Every prospector just needed tools, whatever gold they found.

NVIDIA acts as the tool merchant selling GPUs to anyone building AI capabilities. Cloud providers like AWS, Azure and Google Cloud profit from every AI query and model update, creating recurring revenue streams. The AI data centre gold rush driven by thousands of newcomers has created sustained demand for infrastructure, not just one-time product sales.

The Pattern Every Investor Should Know

The pick-and-shovel play relies on derived demand. Infrastructure suppliers enjoy steadier and more predictable returns than companies competing directly in volatile markets when an industry explodes. Your investment success often depends on recognising this pattern early, then positioning yourself with the companies providing the foundation rather than chasing the most hyped applications.

AI Startups: Burning Cash Despite the Hype

OpenAI's Revenue vs Spending Reality

OpenAI's financials reveal staggering losses behind the headlines. The company reported EUR 12.47 billion in revenue for 2025, but total costs and expenses hit EUR 32.44 billion. This resulted in an operating loss of EUR 19.96 billion. Research and development alone consumed EUR 18.30 billion. OpenAI's net loss attributable to the company reached EUR 36.74 billion when factoring in restructuring from non-profit to for-profit status.

The expense breakdown exposes the financial strain. The cost of revenue amounted to EUR 7.16 billion, sales and marketing to EUR 5.47 billion, and general administrative costs to EUR 1.50 billion. Microsoft received EUR 16.41 billion in total payments from OpenAI during 2025. In 2024, OpenAI spent EUR 2.26 to generate every EUR 0.95 in revenue.

Anthropic's Funding and Cash Burn Rate

Anthropic closed a Series H funding round worth EUR 62.02 billion in April 2026. The round valued the company at EUR 920.81 billion post-money. Run-rate revenue crossed EUR 44.85 billion earlier that month. The company has raised EUR 125.96 billion across 18 funding rounds.

Cash burn remains substantial. Anthropic projected roughly EUR 8.59 billion in annualised revenue for 2025 while expecting about EUR 4.96 billion in cash burn. Inference costs from Google and Amazon servers ran 23% higher than predicted.

Why Most AI Companies Aren't Profitable Yet

Training costs create massive upfront expenses. GPT-4 required an estimated EUR 74.43 million in compute resources, while Google's Gemini Ultra cost EUR 182.25 million. An MIT study found that 95% of organisations implementing AI reported no measurable return on investment.

The Problem with AI Business Models

Inference costs alone factored in 50% of OpenAI's 2024 revenues, while training costs absorbed another 75%. This cost structure exceeds 100% of revenue before other business expenses. Perplexity spent 164% of its 2024 revenue on computing costs, while Cursor allocates 100% of subscription revenue to Anthropic for model access.

The Real Winners: Cloud Providers and Chip Makers

NVIDIA: The AI Gold Rush Stock Leader

NVIDIA reported record revenue of EUR 77.86 billion in Q1 2026, up 85% year-over-year. Data centre revenue reached EUR 71.76 billion, climbing 92% from the previous year. The company's three-month revenue to October jumped 62% to EUR 54.39 billion. Chip demand from AI data centres pushed the figure up. Fourth-quarter sales forecasts of EUR 62.02 billion topped estimates, with CEO Jensen Huang stating that cloud GPUs are "sold out".

Amazon Web Services (AWS) Revenue Surge

AWS hit EUR 35.87 billion in Q1 2026, a 28% growth and the fastest expansion in 15 quarters. The company's AI services generate over EUR 14.31 billion in annualised revenue, accounting for about 10% of AWS's total revenue run rate. Bedrock token processing surged 170% quarter-over-quarter. AWS now operates on a EUR 143.13 billion annual revenue run rate.

Microsoft Azure's AI Infrastructure Dominance

Azure and other cloud services' revenue increased 40% in Q1 2026. Microsoft's AI business surpassed an annual revenue run rate of EUR 35.31 billion, up 123% year-over-year. Microsoft Cloud revenue reached €52.00 billion and grew 29%. Azure annual revenue now exceeds EUR 71.57 billion, with AI services contributing 19% of quarterly growth.

Google Cloud's Profitable Transformation

Google Cloud revenue surged 35% to EUR 10.88 billion in Q3 2024. Google Cloud generated EUR 58.7 billion for 2025, a 36% increase. Operating income soared 154% to EUR 5.06 billion. BigQuery machine learning operations increased 80% over two quarters.

The AI Data Center Gold Rush Driven by Thousands of Newcomers

Capital expenditure from the 14 largest data centre operators reaches EUR 715.66 billion this year against EUR 429.39 billion last year. Over 23 gigawatts of data centre capacity were under construction as of September 2025. McKinsey estimates that companies will invest almost EUR 6.68 trillion in data centre infrastructure by 2030.

What Smart Investors Need to Know

Why Timing Matters More Than You Think

Market concentration has reached alarming levels.   The 10 largest companies in the S&P 500 account for about 40% of the index's total market capitalisation, which is well above the 27% peak during the 1999-2000 tech bubble. Jeremy Grantham, who predicted the dot-com crash, now warns the AI bubble is about to burst. Ray Dalio sees striking similarities to that earlier collapse.

The Bank of England warns that AI company valuations could fall if infrastructure costs prove too high. Investors aren't cautioned about crash risks should AI fall short of expectations.

The Risk of Chasing Performance

Performance chasing leads you into crowded trades vulnerable to overvaluation. Recency bias makes you believe recent rallies will continue indefinitely. Herd mentality drives you to follow others into small-cap funds that don't match your risk tolerance. Past wins create overconfidence and encourage excessive risk-taking.

One analysis found performance chasing returned 154.7% over 10 years but came with calendar year swings from negative 14.2% to positive 44.4%.

Building a Balanced Approach to AI Investments

Vary your holdings throughout the AI value chain rather than concentrating on a few names. Recent headlines may have prompted you to think about your portfolio, or you would like reassurance that your investment strategy remains arranged with your long-term goals. We are always happy to have that conversation. You can contact Expat Fiduciary here.

Regular rebalancing helps manage single-security risk when positions exceed 8-10% of portfolio value. Equal-weighted strategies or mid-cap exposure reduces concentration in the AI gold rush stock leaders.

Final Thoughts

History shows that infrastructure providers win gold rushes while prospectors struggle. The AI boom follows this pattern, with cloud giants and chip makers capturing profits while headline-grabbing startups burn through billions. Your investment success depends on recognising this dynamic before valuations correct themselves.

If recent headlines have prompted you to think about your portfolio or you would like reassurance that your investment strategy remains aligned with your long-term goals, we are always happy to have that conversation. You can contact Expat Fiduciary here.  Vary across the value chain rather than chasing performance.

FAQs

Q1. Who is really profiting from the AI boom?

The primary beneficiaries are infrastructure providers rather than AI startups. Cloud computing giants like AWS, Microsoft Azure, and Google Cloud are experiencing massive revenue growth, with AWS hitting EUR 35.87 billion in Q1 2026. NVIDIA, the chip manufacturer, reported record revenue of EUR 77.86 billion in the same quarter. Meanwhile, high-profile AI companies like OpenAI are operating at significant losses, spending EUR 32.44 billion against EUR 12.47 billion in revenue.

Q2. Why aren't AI startups profitable despite all the hype?

AI companies face enormous infrastructure costs that exceed their revenues. Training advanced models costs tens to hundreds of millions of euros, while inference costs consume 50% or more of revenue. In 2024, OpenAI spent EUR 2.26 to generate every EUR 0.95 in revenue. Additionally, 95% of organisations implementing AI report no measurable return on investment, highlighting the gap between expectations and financial reality.

Q3. How does the AI boom compare to historical gold rushes?

The pattern mirrors the California Gold Rush of 1848, where merchants selling picks, shovels, and supplies became wealthy while most prospectors failed. Samuel Brannan and Levi Strauss built fortunes by providing tools rather than mining gold. Similarly, companies selling AI infrastructure—GPUs, cloud computing, and data centre services—are capturing steady profits while AI application developers struggle with profitability.

Q4. What risks should investors consider with AI investments?

Market concentration has reached concerning levels, with the 10 largest S&P 500 companies representing 40% of total market capitalisation—exceeding the peak of the 1999-2000 tech bubble. Performance chasing can lead to overvalued positions with extreme volatility. The Bank of England warns that AI valuations could fall sharply if infrastructure costs remain too high or if AI fails to meet inflated expectations.

Q5. What investment approach makes sense for the AI sector?

Diversification across the AI value chain reduces risk compared to concentrating in hyped startups.   Focus on infrastructure providers with proven revenue streams rather than chasing performance in unprofitable AI applications. Regular portfolio rebalancing helps manage concentration risk when individual positions exceed 8-10% of total value. Equal-weighted strategies can reduce overexposure to the largest AI-related stocks.