Interest Rates and Earnings: Two Tests of the Same AI Thesis

Higher rates can affect:

• Corporate borrowing.
• Project finance.
• Data-center development.
• Venture funding.
• Customer technology budgets.
• Equity valuation.
• Acquisition activity.

The policy rate does not determine every financing cost directly, but it influences the broader environment in which risk is priced.

A credible AI strategy must remain economically defensible across more than one interest-rate scenario.

Earnings must connect spending to operating outcomes

Investors need more than capital-expenditure guidance. They need evidence explaining what the spending enables.

Useful disclosures include:

• AI-related revenue.
• Cloud demand.
• Capacity utilization.
• Customer conversion.
• Cost per workload.
• Margin impact.
• Product retention.
• Internal productivity.

Management should distinguish revenue generated by AI products from general growth that coincides with AI investment.

The relationship may be indirect, but the mechanism should be clear. For example, an AI feature may improve advertising performance, reduce customer churn, or increase cloud consumption.

Capital expenditure creates a cash-flow test

A profitable company can experience pressure when capital spending grows faster than operating cash flow.

Investors should examine:

• Free cash flow after capital expenditure.
• Hardware-refresh requirements.
• Lease and purchase commitments.
• Debt issuance.
• Supplier financing.
• Power contracts.
• Depreciation.

The initial buildout may support several years of growth, or it may require continuous reinvestment to remain competitive.

The difference determines whether AI infrastructure becomes an asset that compounds value or a permanent claim on cash generation.

Demand quality matters more than enthusiasm

AI usage can grow rapidly because products are free, bundled, subsidized, or included in existing contracts. Usage alone does not prove willingness to pay.

Earnings analysis should examine:

• Paid versus free activity.
• Contract duration.
• Customer concentration.
• Expansion behavior.
• Renewal patterns.
• Price realization.
• Workload profitability.

A customer experimenting with several models is different from one embedding a provider into a critical workflow.

Durable demand is reflected in operational dependence, measurable value, and continued payment rather than temporary curiosity.

Margins reveal where value is captured

AI can create revenue while shifting value toward chip suppliers, energy providers, data-center operators, or model vendors.

A software company may sell an AI feature but pay substantial inference costs. A cloud provider may increase revenue while financing expensive capacity. A chip supplier may capture strong margins while its customers face uncertain monetization.

Investors should map value across the chain:

1. Who supplies capital?
2. Who controls scarce infrastructure?
3. Who owns the customer relationship?
4. Who bears usage cost?
5. Who can raise prices?
6. Who absorbs obsolescence?

The strongest revenue growth does not automatically identify the strongest return on capital.

Management guidance should disclose assumptions

Forecasts become useful when investors understand the assumptions behind them.

Relevant assumptions include:

• Customer growth.
• Utilization.
• Model cost.
• Hardware availability.
• Energy pricing.
• Product conversion.
• Capital intensity.
• Competitive pricing.

Management should explain which assumptions are most uncertain and what would cause spending to accelerate or slow.

A forecast that depends on permanently falling inference cost and permanently rising demand should be stress-tested against less favorable combinations.

The Fed can influence customer behavior indirectly

The effect of monetary policy extends beyond Big Tech financing. Higher borrowing costs can reduce technology spending among startups, enterprises, consumers, and public institutions.

Customers may delay migrations, negotiate harder, reduce experimental budgets, or consolidate vendors. This can affect AI revenue even when the technology remains strategically important.

Companies serving capital-sensitive customers should monitor whether usage growth translates into paid expansion under tighter budgets.

AI products that demonstrate direct cost savings or revenue improvement may remain more resilient than tools justified primarily by experimentation.

Investors need scenario-based valuation

One forecast cannot capture the uncertainty around rates, capital spending, competition, and monetization.

A useful valuation framework may consider:

• Stable rates and strong AI monetization.
• Higher rates with strong demand.
• Falling rates with weak utilization.
• Price competition and lower model margins.
• Infrastructure delays.
• Faster efficiency gains that reduce capital needs.

Each scenario should connect operational assumptions to cash flow and required return.

The objective is not to predict the exact central-bank decision. It is to understand which companies remain attractive when financing and demand assumptions change.

Boards should preserve spending flexibility

AI infrastructure decisions can involve long contracts, capacity reservations, and construction commitments. These reduce the ability to respond quickly to weaker demand.

Boards should ask:

• Which spending is reversible?
• Which commitments are fixed?
• Can capacity serve several products?
• Can workloads move between providers?
• Which projects depend on one customer?
• What triggers a reassessment?

Flexible architecture and staged investment can preserve strategic participation without committing the entire capital plan to one forecast.

The test is economic, not rhetorical

The market does not need proof that AI can produce impressive outputs. It needs proof that companies can convert those outputs into sustainable financial returns while paying for infrastructure and capital.

Federal Reserve policy changes the hurdle rate. Earnings show whether companies are clearing it.

The strongest businesses will demonstrate disciplined spending, identifiable customer value, improving unit economics, and enough flexibility to adapt when interest rates, competition, or technology move differently from the original plan.