What Are Companies Getting for All That A.I. Spending?

Companies are spending billions on AI infrastructure without clear ROI metrics. The emerging field of 'tokenomics' attempts to quantify returns per dollar invested—a real-world example of capital budgeting discipline and measuring whether spending decisions actually create value.

Read article

Teaching notes are auto-generated. Worth a fact-check before class.

Why this matters

Companies like Microsoft, Google, and Meta are investing tens of billions annually in artificial intelligence—buying specialized computer chips, building data centers, and hiring AI researchers. Unlike a typical factory machine you can watch produce widgets, AI's payoff is abstract: it might improve worker productivity, enable new products, or optimize existing operations. The problem: executives can't easily prove these investments are worth the cost. A new framework called 'tokenomics' tries to measure this by tracking how much additional revenue or profit each dollar of AI spending generates. This is capital budgeting in the real world—deciding whether today's huge cash outflow will produce enough future cash inflows to justify the expense.

Key points

AI spending hit record levels without proven returns, forcing companies to develop new metrics like tokenomics → capital investments must be justified by measurable future cash flows, not just technology hype.
Tokenomics tracks revenue or profit per dollar invested in AI infrastructure → similar to how you'd evaluate whether a $2 million manufacturing machine will eventually pay for itself.
Companies struggle to isolate AI's impact on margins and efficiency ratios → management effectiveness depends on choosing investments that actually improve profitability and asset utilization.
Time lag between AI spending and measurable results creates uncertainty → net present value calculations must account for the risk that benefits may arrive later or smaller than promised.

Key terms

Capital investment: A large, long-term purchase (like a building, equipment, or AI infrastructure) expected to generate cash or profit for multiple years.
Net present value (NPV): A calculation that compares the total value of all future cash a project will earn (adjusted for the time value of money) against its upfront cost; a positive NPV means the investment is worth undertaking.
Time value of money: The principle that cash you receive today is worth more than the same amount received in the future because you can invest it and earn returns.
Return on investment (ROI): A ratio measuring how much profit or benefit you earn from every dollar spent; calculated by dividing profit gained by total cost.
Asset turnover ratio: A measure of how efficiently a company uses its assets to generate sales, calculated by dividing total revenue by average total assets.
Operating margin: A profitability ratio showing what percentage of each sales dollar becomes operating profit after paying all day-to-day expenses.
Tokenomics: An emerging framework that quantifies the financial return from artificial intelligence investments, typically by measuring additional revenue or profit generated per dollar spent on AI infrastructure.

Discussion prompts

01
Why is betting that 'AI is good for business' not the same as proving an AI investment has a positive net present value?
Answer: Students should recognize that 'technology is valuable' is a belief, not proof. Net present value requires concrete cash flow forecasts, a discount rate, and honest comparison to the cost. Without these, you're guessing, not investing. The key is to push them to name what specific metric (revenue, profit, efficiency) will prove the AI worked.
02
If Microsoft's AI spending takes three years to show up in profit margins, how should that timeline affect today's investment decision?
Answer: Time matters because of the time value of money: a dollar earned in three years is worth less than a dollar earned today (you could have invested today's dollar and earned interest). If the three-year timeline makes the NPV negative or barely positive, the investment isn't worth the risk. Use a simple example: $1 earned in year 3 at 10% discount rate is worth ~$0.75 today, not $1.
03
Would you rather invest $1 billion in AI with a guaranteed 10% annual return, or $500 million in AI with an uncertain 25% return? Why?
Answer: Neither is objectively right—it depends on the company's risk tolerance and cost of capital. The stronger answer acknowledges the trade-off: guaranteed 10% is safe and predictable (good for conservative firms); uncertain 25% is higher reward but riskier (only good if the company can afford to lose that $500M). Draw out that capital budgeting is ultimately about fitting investment choices to the company's risk appetite.
04
What efficiency ratio would best show whether a company's AI spending actually improved how well management runs the business?
Answer: Asset turnover ratio (revenue ÷ assets) or operating margin (operating profit ÷ revenue) would both reveal whether AI improved efficiency. If neither ratio improves after the AI spend, management either bought the wrong asset or overestimated payoff. Push students to see that ratios translate abstract 'AI value' into concrete, measurable management performance.

Bringing it to class

Start by asking students to imagine their parents spent $50,000 on a new piece of equipment for a family business without any plan to measure whether it paid off. That's what's happening at scale. Then draw a simple timeline on the board: Year 0 (AI cost: –$1B), Year 1–3 (uncertain profit gains). Ask: 'How do we decide if this is worth it?' Lead them to NPV—it's the only fair way to compare today's certainty (the cost) with tomorrow's uncertainty (the benefits). Use the tokenomics concept as the real-world attempt to measure what companies are still struggling to prove. End by pointing out that without discipline (NPV, payback period, ROI targets), management is just spending—not investing.