The Core Economic Flaw in the AI Bull Case

Artificial intelligence remains one of the market’s most powerful investment themes. Chipmakers, cloud providers, data-center operators and software companies have attracted substantial capital as investors anticipate rapid adoption and long-term productivity gains.

The concern identified by Apollo chief economist Torsten Slok is not that artificial intelligence lacks value. It is that the sector’s apparent profitability may depend too heavily on investor funding rather than durable payments from end customers. Companies could report strong growth while the broader ecosystem still requires continuous capital to sustain its expansion. Source 9

That distinction matters as stocks and bonds weaken together. Higher bond yields can reduce the value investors assign to distant earnings, increase infrastructure-financing costs and expose companies whose valuations assume years of exceptional growth.

The market reversal does not prove that artificial intelligence has failed. It does, however, create a test: can AI companies generate enough cash from customers to justify the capital committed to the industry?

What Is the Core Economic Flaw in the AI Bull Case?

Investor-Funded Growth Versus Customer-Funded Growth

Investor-funded growth occurs when expansion depends mainly on venture capital, public equity, strategic investments or debt. This capital can help a company build products, hire employees and construct infrastructure before demand fully develops.

Customer-funded growth works differently. Customers pay enough for a company’s products or services to cover operating expenses, infrastructure investment and a reasonable return on capital. The company expands because its commercial activity creates cash, not because new investors must continually finance losses or aggressive spending.

Slok’s criticism focuses on the possibility that AI companies are being supported more by investors than by customers. Capital flows through the sector: investors fund AI companies, AI companies purchase hardware and cloud capacity, and suppliers report rising demand. Those results then reinforce investor confidence.

This cycle is not automatically fraudulent. Companies often invest ahead of demand, and new industries frequently require substantial funding before becoming profitable. The issue is durability. If customers are only testing AI, accepting subsidized pricing or shifting existing technology budgets, reported growth may not represent a permanent expansion in economic value.

The key question is whether AI companies can maintain growth when investors become less willing to finance aggressive expansion.

Why AI Requires Extraordinary Capital Spending

Scaling advanced AI requires far more than software development. The industry needs:

• Advanced processors and servers.
• Large data centers.
• Electricity generation and grid capacity.
• Cloud infrastructure and networking equipment.
• Specialized engineers and researchers.
• Cooling systems, security controls and compliance infrastructure.

These costs raise the industry’s break-even point. An AI provider may report rapidly rising revenue while still requiring substantial capital expenditure to serve customers. Hardware must be purchased, facilities must be built, and models must be trained and maintained.

Investment ahead of demand can be economically rational if future usage expands enough to absorb the capacity. The risk arises when construction outpaces profitable demand. A company can then face underused data centers, falling prices and high depreciation expenses.

Revenue growth alone does not resolve this problem. Investors must assess whether each dollar of revenue produces enough gross profit and free cash flow to support the next phase of investment.

How Stock and Bond Trades Are Unraveling

Why AI Stocks Are Sensitive to Interest Rates

High-growth technology stocks are especially sensitive to interest rates because much of their valuation depends on expected future earnings. When bond yields rise, those future cash flows are discounted more heavily, reducing the present value of earnings expected many years from now.

Higher yields also increase borrowing costs. That matters for AI businesses because the sector requires substantial investment in data centers, processors, energy and networking. Companies that depend on debt or frequent equity issuance face more expensive capital when financial conditions tighten.

The market can therefore reassess AI-related shares even when their operating results remain strong. Investors may conclude that a company’s growth is valuable but not valuable enough to justify its current price. They may also question whether projected margins will emerge after accounting for computing expenses, hardware depreciation and energy costs.

Why Stocks and Bonds Moving Together Matters

Traditional portfolios often rely on stocks and bonds to respond differently to market stress. Bonds may cushion equity losses when investors seek safety. That diversification becomes less effective when both asset classes decline simultaneously.

A simultaneous stock-and-bond selloff can produce several effects:

1. Investors reduce exposure to expensive growth stocks.
2. Higher yields lower the value of long-duration assets.
3. Borrowing costs rise for infrastructure projects.
4. Private-market funding becomes more selective.
5. Companies face greater pressure to produce near-term cash flow.

This combination is important for AI because the sector depends on investor confidence, high valuation multiples and accessible capital. If those conditions weaken together, companies must rely more heavily on actual customer payments.

A Market Correction Is a Stress Test, Not Proof of Failure

Market weakness is not definitive evidence that AI is unproductive. A valuable technology can experience a speculative investment cycle. A correction may instead reveal which companies have durable economics and which depend on optimistic assumptions.

Investors are likely to examine:

• Customer retention and renewal rates.
• Expansion revenue from existing customers.
• Contract duration and pricing.
• Free cash flow.
• Capital-expenditure intensity.
• Computing costs per customer.
• Return on invested capital.
• Measurable productivity gains.

The strongest businesses should show that customers continue to spend because AI creates savings, new revenue or strategic advantages. Weaker businesses may reveal that adoption depends on discounts, subsidies or fear of falling behind competitors.

The Circular Economics Concern in AI

How AI Companies May Support One Another

The AI ecosystem can create a reinforcing loop:

1. Investors provide capital to AI companies.
2. AI companies purchase chips, cloud services and data-center capacity.
3. Suppliers report strong demand and expand production.
4. Strong supplier results reinforce investor enthusiasm.
5. Higher valuations make it easier for companies to raise more capital.

This loop can support genuine innovation. It can also obscure the final source of demand. If end customers do not generate enough value from AI, the system remains dependent on new investment.

The concern is especially relevant when companies within the same ecosystem buy from one another. A chip supplier may sell to a cloud provider, which rents capacity to an AI developer, which sells services to enterprises. Each transaction can be legitimate, but the economic test remains whether outside customers ultimately pay enough to support the chain.

Investors should ask whether spending reflects measurable productivity improvements, competitive pressure or speculative expectations about future demand.

Revenue Does Not Always Equal Economic Value

AI revenue can grow without producing equivalent economic value for customers. Several factors can weaken revenue quality:

• Customers may be paying for short-term pilots rather than permanent deployments.
• AI budgets may replace spending on existing software rather than create new output.
• Providers may offer discounts, credits or free usage to accelerate adoption.
• Customers may tolerate early losses while searching for practical applications.
• High computing costs may limit the provider’s eventual margins.

More useful indicators include retention, expansion revenue, gross margins after computing costs, usage-based profitability and customer payback periods.

A bank may spend millions on an AI system, for example, but the relevant question is not merely whether the AI vendor records revenue. The relevant question is whether the bank reduces costs, increases revenue, improves risk management or processes more work with the same resources.

Evidence Supporting the AI Bull Case

AI Revenue Is Expanding in Major Technology Businesses

The bullish case has substantial evidence. One Fortune report said SpaceX posted its first operating profit as a public company and that its AI-related revenue nearly tripled to $7.6 billion, surpassing the reported contribution from Starlink. Source 3

If accurately defined and reported, that growth would demonstrate that AI demand can become a significant commercial business rather than remain a research expense. It would also show that AI-related revenue can exceed established technology businesses in some corporate structures.

The figures require careful interpretation. Revenue growth does not establish long-term profitability, and one company’s results cannot prove that the entire AI sector has sound economics. Investors must confirm what the company classifies as AI revenue, how operating profit is calculated and whether the figures include related businesses or internal transactions.

Still, strong commercial results provide meaningful evidence against the claim that AI demand is purely speculative.

AI Adoption Is Expanding Across Industries

The AI economy is also spreading into financial services and automated commerce. Mural, a startup founded by former Palantir engineers, reportedly raised a $15 million bid to connect banks with the emerging “bot economy.” Its objective is to help financial institutions interact with AI agents and automated commerce. Source 5

This trend suggests that AI agents may move beyond chat interfaces into payments, banking operations and automated purchasing. Financial institutions are preparing for systems that can communicate, negotiate and execute tasks with limited human intervention.

The unresolved issue is commercial viability. Infrastructure for future demand does not guarantee profitable demand. Investors must identify who pays, how much they pay and what economic result the service produces.

Unequal Access Shows Both Demand and Risk

Demand for AI-powered education provides another example of willingness to pay. Fortune reported that wealthy parents were paying as much as $75,000 for AI-powered schooling designed with guardrails, amid concerns about literacy and traditional school performance. Source 7

The example indicates demand for personalized learning and advanced educational tools. It also highlights unequal access. High prices can demonstrate scarcity and willingness to pay, but they can concentrate benefits among families with substantial financial resources.

This evidence does not directly prove or disprove the profitability of AI infrastructure companies. It shows that certain users value AI applications enough to pay premium prices. The broader economic question is whether those applications can scale beyond affluent customers while producing measurable educational gains.

What the Market Could Be Pricing Incorrectly

Demand May Not Automatically Catch Up With Capacity

The bullish assumption is that companies are building infrastructure ahead of demand. As AI adoption expands, customers will absorb the capacity, providers will gain pricing power and early investment will generate high returns.

The risk is that capacity may be built faster than profitable use cases emerge. Competition could drive prices down, particularly if AI services become standardized. Customers may also negotiate lower prices as multiple providers offer similar models and computing capacity.

Revenue Growth May Not Guarantee Strong Margins

AI services carry unusual costs:

• Computing and electricity.
• Hardware depreciation.
• Model training and maintenance.
• Data acquisition and storage.
• Security and regulatory compliance.
• Human oversight and customer support.

A provider can increase revenue while free cash flow remains weak. Investors should therefore focus on unit economics, pricing power and returns after infrastructure costs.

Capital May Not Remain Available

High valuations make it easier for companies to raise equity. Strong credit markets support data-center construction, while strategic investors can finance suppliers and customers. This capital helps sustain the AI investment cycle.

If sentiment changes, funding rounds may become smaller or more expensive. Private valuations may decline, data-center projects may be delayed and public shareholders may demand profits sooner.

That shift would expose companies with weak customer economics. Businesses with recurring revenue and strong cash generation could continue investing. Companies dependent on external financing would face greater pressure to reduce spending or raise capital under unfavorable terms.

How Investors Can Evaluate the AI Trade

Questions About the Customer

Investors should ask:

• Who is paying for the AI product?
• Does the customer renew and expand usage?
• Does the product create measurable savings or revenue?
• Can the customer remain profitable after paying for AI services?
• Is adoption driven by business value or fear of being left behind?

Questions About the Company

Important company-level questions include:

• Does the business generate free cash flow?
• Are margins improving after computing costs?
• How much capital expenditure supports each dollar of revenue?
• Does the company depend on related-party transactions or strategic subsidies?
• Can it maintain growth if funding conditions deteriorate?

Questions About Valuation

Investors should also examine:

• How much future growth is already reflected in the share price?
• What happens if revenue growth slows?
• How sensitive is the valuation to bond yields?
• Does the company need perfect execution to justify its market capitalization?
• Are investors paying for current earnings or speculative future dominance?

AI Can Be Transformative Without Making Every AI Investment Attractive

Technology quality and investment returns are separate questions. Artificial intelligence may transform productivity, education, finance and commerce while some AI companies deliver poor shareholder returns.

Earlier infrastructure booms show why. New technologies can create enormous social value, but suppliers and investors may overpay during periods of rapid expansion. Competition can eventually push prices down, while high capital requirements prevent companies from earning adequate returns.

Potential winners may include companies with:

• Proprietary distribution.
• Recurring customer revenue.
• Efficient computing infrastructure.
• Strong customer retention.
• Products linked to measurable business outcomes.
• Pricing power and positive free cash flow.

Vulnerable companies may include businesses that:

• Depend on external financing.
• Carry high operating costs.
• Lack pricing power.
• Sell mainly to other AI companies.
• Have unclear customer demand.
• Require perpetual growth to justify their valuations.

The technology can succeed while individual companies fail. AI adoption can increase while investors lose money in overvalued stocks.

Conclusion: The Core Test Is End-Customer Economics

The central criticism of the AI bull case is that investor capital may be supporting the sector’s apparent profitability more than customer payments are. If that is true, the industry remains vulnerable when markets demand cash flow, lower risk and credible returns on capital.

Weak stocks and bonds can expose this dependence. Higher yields reduce the value of distant earnings, increase financing costs and pressure companies whose valuations assume years of rapid growth. A market correction does not invalidate AI’s potential, but it makes the sector’s financial structure easier to examine.

There is meaningful evidence that AI demand is real. Revenue is expanding in major technology businesses, financial institutions are preparing for AI agents and users are paying for premium applications. Yet adoption, revenue and productivity are different measures.

The decisive question is whether AI companies can generate durable profits from customers after accounting for computing, infrastructure, energy and capital costs. The next phase of the AI trade will depend less on excitement about potential and more on evidence of repeatable, customer-funded returns.

Frequently Asked Questions

What is the core economic flaw in the AI bull case?

The concern is that AI companies may appear profitable because investors are funding their expansion, while customers are not yet generating enough revenue or savings to support the ecosystem independently.

Why do higher bond yields pressure AI stocks?

Higher yields reduce the present value of expected future earnings and increase financing costs for companies that require substantial capital expenditure.

Does rapid AI revenue growth prove that the sector is sustainable?

No. Revenue growth indicates demand, but sustainability also requires durable customer spending, healthy margins, free cash flow and returns above the cost of capital.

What does customer-funded growth mean?

Customer-funded growth means customers pay enough for a company’s products to cover operating costs, infrastructure investment and an appropriate return without continuous dependence on new investor capital.

Could AI transform the economy if some AI stocks decline?

Yes. Technological impact and investment performance are separate. AI may produce significant productivity gains while individual companies or overvalued stocks generate poor returns.

What should investors watch next?

Investors should monitor customer retention, usage economics, free cash flow, computing costs, capital expenditure, pricing power and evidence that AI deployments deliver measurable business results.