AI Can Create Real Value and Still Be Overvalued

AI adoption, revenue and productivity gains are becoming increasingly measurable, but the scale of capital flowing into AI is raising a harder question: whether future returns can justify today’s valuations and infrastructure spending.

Artificial intelligence is no longer a technology story built primarily on future promises. Companies are generating substantial AI-related revenue, enterprises are deploying the technology at scale, and measurable productivity gains are emerging across several business functions. Stanford’s 2026 AI Index found that 88% of surveyed organizations used AI in at least one business function in 2025, while generative AI reached 53% adoption within three years. The same research found productivity gains of 14% to 15% in customer support, 26% in software development and 50% in marketing output in the studies it reviewed.

Yet the existence of real economic value does not automatically validate every AI valuation. Private AI investment grew 127.5% in 2025, while generative AI investment more than doubled. At the same time, compute costs and infrastructure spending have reached record levels. The question facing the market is therefore becoming more complicated than whether AI works. It is whether the economic value eventually generated by AI will be large enough, and arrive quickly enough, to justify the capital being committed today.

AI’s Value Is Becoming Easier to Measure

There is growing evidence that AI is producing tangible economic benefits. The IMF estimates that technology investment related to AI contributed around 0.5 percentage points to US GDP growth in 2025, while noting that productivity growth has accelerated in recent years and may partly reflect AI adoption. Enterprise adoption is also expanding. Microsoft reported $90 billion in quarterly revenue for the period ended June 30, 2026, while Azure revenue surpassed $100 billion for the first time and Microsoft 365 Copilot reached more than 30 million paid seats. Microsoft also reported that nearly 90% of its annual cloud revenue came from customers outside frontier AI model companies, suggesting that demand is increasingly extending beyond the AI labs themselves.

These developments matter because they establish an important distinction: AI can be economically valuable even if some AI assets are financially overvalued.

The Market Is Pricing Future AI Economics

Valuation, however, depends not only on what a company earns today, but on what investors expect it to earn in the future. That distinction is becoming increasingly important as private AI companies command extraordinary valuations and require enormous amounts of capital to support their growth. INSEAD researchers argue that a bubble occurs when prices exceed what future fundamentals can realistically deliver, rather than simply when prices appear high relative to current earnings. They also note that many leading AI infrastructure companies have genuine earnings growth, making the current situation more nuanced than the dot-com era.

For investors, the critical question is therefore whether expected growth is supported by measurable commercial performance or increasingly by the AI narrative itself. Urska Vracun, an angel investor, raised this distinction while conversing with AsiaTechDaily.

“I do think that AI market is vulnerable to a significant correction and investment portfolios which focus solely on AI-driven startups can expect many things. The market displays many attributes of a bubble although the technology itself has created real value. For me, the clearest warning sign would be when companies are valued primarily on their association with AI rather than on measurable revenue, margins, customer adoption or a credible path to profitability. Another signal would be excessive spending on AI infrastructure without sufficient evidence that the resulting computing capacity will generate adequate returns.”

Her point shifts the debate from whether AI is real to whether investors are accurately pricing the economics of individual businesses.

The scale of AI infrastructure investment makes that question harder to ignore. Microsoft spent $41 billion on capital expenditures in its latest quarter, with roughly two-thirds directed toward short-lived assets, primarily CPUs and GPUs. The company expects more than $50 billion in capital expenditure in the following quarter.

Amazon‘s capital intensity is also rising. Its trailing 12-month purchases of property and equipment reached $169 billion by June 2026, up 64% year over year, while AWS revenue grew 37% in the second quarter. Meta expects 2026 capital expenditure of $125 billion to $145 billion, including investment to support future data-center capacity.

These investments are not evidence of irrational spending by themselves. Demand for AI compute is clearly growing. The harder question is whether revenue and cash generation can ultimately grow fast enough to produce attractive returns on the infrastructure being built. That challenge is particularly visible at AI model companies. Reuters reported in September that Anthropic has disclosed at least $518 billion in long-term AI infrastructure commitments over the next decade, with approximately 80% of those commitments binding regardless of usage. The market therefore has to evaluate not simply how much AI infrastructure is required, but how much of that capacity can be converted into sustainable economic output.

The Next Phase Will Be About Returns

The valuation question is increasingly relevant across Asia as well. Asian startups raised $42.8 billion in the second quarter of 2026, according to Crunchbase, the strongest quarterly total in more than three years. More than 60% of that funding went to AI startups, while the number of deals fell to a multiyear low, indicating that capital was becoming concentrated among fewer companies.

KPMG separately recorded $50.8 billion across 2,676 Asian venture deals during the quarter, with AI, robotics, semiconductors, infrastructure and advanced manufacturing among the major investment themes. The numbers show strong investor conviction, but they also raise a broader question for the region: when AI becomes one of the dominant destinations for venture capital, how can investors distinguish businesses with durable economics from companies benefiting primarily from the strength of the category?

AI does not need to fail for AI investments to disappoint. A company can build a genuinely valuable technology and still struggle to justify its valuation if revenue growth slows, margins remain weak, customer adoption does not scale as expected or the capital required to support growth becomes excessive. That is why the next phase of the AI market may be less about proving that artificial intelligence works and more about proving who can make sustainable money from it.

For investors, that means looking beyond AI exposure toward revenue quality, margins, customer retention, adoption and cash generation. For AI companies, it means demonstrating that technological capability can translate into durable commercial value. The technology may ultimately transform industries. The financial question is whether today’s prices are correctly anticipating how much of that transformation will become profitable, and which companies will capture it.

Quick Takeaways

• AI is creating real economic value, with rising enterprise adoption, revenue and measurable productivity gains.
• Real value does not automatically justify high valuations. Investors still need evidence of revenue growth, margins, customer adoption and a credible path to profitability.
• AI infrastructure spending is becoming a major risk factor, as hyperscalers commit hundreds of billions of dollars to GPUs, data centers and computing capacity.
• Asia is participating heavily in the AI investment cycle, with more than 60% of Asian startup funding in Q2 2026 going to AI startups.
• Capital is becoming increasingly concentrated, raising questions about whether investors are funding durable businesses or simply following the strongest AI narrative.
• A correction would not necessarily mean AI has failed. AI could continue transforming industries while individual companies or investments prove overvalued.
• The next test for the AI market is returns: whether today’s enormous investment can ultimately translate into sustainable revenue, margins and cash generation.