Compare AI Companies: Revenue Margins & Infrastructure
This guide shows you how to compare AI companies by their revenue margins and infrastructure costs. You’ll learn what financial metrics to pull from public filings, how to calculate gross and operating margins, and how to assess infrastructure spending like data centers and compute. By the end, you’ll have a clear framework for evaluating AI businesses side by side.
What You’ll Need
- Access to the companies’ annual reports (10-K for US firms) or quarterly earnings releases
- A spreadsheet (Excel, Google Sheets) for calculations
- Basic understanding of income statement terms (revenue, COGS, operating expenses)
- Company ticker symbols or names
Step-by-Step
- Collect revenue and cost data. For each company, find the income statement in the latest annual report. Note total revenue, cost of goods sold (COGS), and operating expenses. If a company reports “cost of revenue” instead of COGS, use that.
- Calculate gross margin. Gross margin = (Revenue – COGS) / Revenue × 100. This shows how much money is left after direct costs of delivering services. For AI companies, COGS often includes cloud hosting, data center depreciation, and support staff.
- Calculate operating margin. Operating margin = Operating income / Revenue × 100. Operating income is revenue minus COGS and operating expenses (R&D, sales, marketing, G&A). This reflects overall profitability from core operations.
- Identify infrastructure costs. Look for line items like “data center costs,” “server depreciation,” “cloud computing,” or “bandwidth.” These may be in COGS or in a separate note. Also check capital expenditure (capex) on property and equipment, which indicates future infrastructure spending.
- Compute infrastructure intensity. Divide total infrastructure costs (operating + capex) by revenue. This ratio tells you how much of each revenue dollar goes to compute and physical assets. A higher ratio means heavier infrastructure burden.
- Adjust for one-time items. Remove non-recurring charges like restructuring or legal settlements from operating income to get a cleaner picture. Also note any stock-based compensation, which is common in AI startups and can inflate costs.
- Compare across companies. Put the metrics side by side. Look for companies with high gross margins (above 60% is typical for software) but low operating margins, indicating heavy R&D or sales spend. Also compare infrastructure intensity—a company with 30% intensity is more capital-heavy than one at 10%.
- Consider growth stage. Early-stage AI firms often have negative operating margins due to heavy investment. Compare companies at similar maturity levels, or look at trends over 3-5 years to see if margins are improving.
- Check for hidden costs. Some AI companies lease compute from cloud providers, which shows up as operating expense rather than capex. Others build their own data centers, which appears as depreciation. Adjust for these differences to make apples-to-apples comparisons.
- Summarize your findings. Rank companies by margin quality and infrastructure efficiency. A company with 70% gross margin and 20% operating margin is more profitable than one with 50% gross margin and 5% operating margin, even if revenue is similar.
Common Mistakes
- Mixing GAAP and non-GAAP numbers. Always use the same basis (GAAP or adjusted) for all companies. Non-GAAP can exclude stock-based compensation, making margins look better than they are.
- Ignoring depreciation. AI companies often have massive server farms. Depreciation is a real cost that reduces margins—don’t overlook it.
- Comparing a startup to a giant. A $10M revenue startup will have different cost structures than a $10B company. Compare within similar revenue ranges.
- Forgetting R&D costs. AI R&D is huge. If you only look at gross margin, you miss the massive spending on researchers and compute for training models.
- Using only one year. Margins fluctuate. Look at a multi-year trend to see if infrastructure costs are scaling well with revenue.
Tips
- Use the cash flow statement to see actual cash spent on infrastructure, not just accounting depreciation.
- For private companies, look for funding rounds that disclose revenue and costs, or use estimates from reputable industry reports.
- If a company separates “platform costs” from “services costs,” treat them separately—platform costs are more infrastructure-related.
- Check footnotes for cloud computing commitments. Long-term contracts with AWS or Azure can indicate future fixed costs.
- When calculating margins, use revenue net of any customer acquisition costs that are capitalized—this can distort gross margin.
Troubleshooting
Data is missing or inconsistent. If a company doesn’t break out infrastructure costs, estimate using depreciation plus rent and utilities. For cloud costs, look at “cost of revenue” notes. If you can’t find exact numbers, use the ratio of COGS to revenue as a proxy.
Margins look wildly different. This often happens because one company capitalizes software development while another expenses it. Check the accounting policies section to adjust.
You’re comparing different fiscal years. Ensure all data is for the same period (e.g., fiscal 2024). If a company’s fiscal year ends in March, align it with others.
Negative margins confuse you. A negative operating margin isn’t necessarily bad if the company is growing fast and investing. Look at gross margin to see if the core business is profitable before overhead.
Frequently Asked Questions
Q: What is a good gross margin for an AI company?
A: For pure software AI, gross margins are often 60-80%. If they sell hardware or have heavy cloud costs, 40-60% is more typical.
Q: How do I find infrastructure costs in a 10-K?
A: Look in the Management Discussion & Analysis (MD&A) section for “data center” or “infrastructure.” Also check the notes on property, plant, and equipment.
Q: Should I include stock-based compensation in operating expenses?
A: Yes, for a conservative view. It’s a real cost to shareholders, though some companies exclude it in non-GAAP figures.
Q: What if a company doesn’t report COGS separately?
A: Some AI firms report only total operating expenses. In that case, you can’t calculate gross margin—use operating margin instead, and note the limitation.
Q: How does infrastructure cost affect valuation?
A: High infrastructure costs reduce free cash flow, which can lower valuation multiples. Investors often penalize companies with capital-intensive models.
Q: Can I compare an AI chip maker to an AI software company?
A: Not directly—chip makers have manufacturing costs (lower margins), while software has high margins. Compare within the same subsector.
Related
Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.