Reading a 10-K as an Engineer: What to Actually Look For

Engineers who only read earnings calls miss the parts of a business that actually constrain technology investment. A 10-K is slow, repetitive, and full of boilerplate — and buried in Item 1A (Risk Factors) and the MD&A footnotes are signals that change how I prioritize roadmaps, evaluate vendors, and talk to boards.

What I read, in order, in ~90 minutes

I do not read a 10-K cover to cover. I read it like a core dump: specific sections, specific grep patterns.

1. Item 1A — Risk Factors (first pass: diffs)

Public companies revise risk factors every quarter. New paragraphs are the story. I diff the current 10-K against the prior year's filing using plain text export from SEC EDGAR (https://www.sec.gov/cgi-bin/browse-edgar) — no fancy tool required.

Things I searched for in 2024–2025 filings across our vendor and competitor set:

• "artificial intelligence" / "generative AI" — often precedes capex spikes
• "cybersecurity" / "ransomware" — check for incident disclosure language changes
• "supply chain" / "single source" — hardware teams already know; software people miss dependency on one cloud region
• "foreign exchange" / "geographic concentration" — revenue concentration in Greater China affects product and data residency decisions

Example: a SaaS vendor we evaluated added a risk factor about "increased customer scrutiny of subprocessor data transfers" in their FY2024 10-K. Six months later they announced EU data residency pricing. The 10-K beat the press release.

2. Consolidated statements — R&D and capex trajectory

From the income statement and cash flow statement:

Line item	What I infer
R&D as % of revenue	Engineering culture vs margin pressure
R&D absolute growth vs revenue growth	Are they investing ahead of growth or harvesting?
Capitalized software development costs	Aggressive capitalization can inflate earnings
Purchases of property and equipment	Data center buildout vs cloud-native

A company growing revenue 12% YoY but R&D 2% is telling you something about feature velocity for the next two years — regardless of what the CEO said on the call.

3. Segment and geographic revenue

Item 1 and Note disclosures on segments. I map segments to product lines we depend on. If "International" is 8% of revenue but 40% of our user base runs through that stack, support and reliability investment there is a bet against the company's stated priorities.

Geographic concentration also flags sanctions and export control exposure for hardware — relevant if you buy edge devices or SoCs from a vendor with heavy APAC manufacturing.

4. Debt, covenants, and liquidity (Note on long-term debt)

Engineers ignore this until a vendor gets acquired or does layoffs. Look for:

• Current ratio and cash runway implied by quarterly burn (for unprofitable vendors)
• Covenant language on minimum liquidity — triggers restructuring
• Goodwill impairment history — past acquisitions that failed

5. Stock-based compensation and share count

From the equity compensation note and diluted share count. High SBC as % of revenue at a "profitable" vendor means the profit is structurally different from yours if you are bootstrapped or PE-owned.

What I skip on first read

• Auditor report (unless going concern language — then stop everything)
• Most of Item 1 business description (unless new market entry)
• Executive compensation tables (unless governance matters for partnership)

How this connects to engineering decisions

Vendor selection: A 10-K will not replace a technical eval, but it tells you whether their cloud egress story (see honest egress accounting) matches their margin structure. Low gross margin + high egress-dependent architecture = future price increases on your bill.

Build vs buy: If a public competitor's R&D line is flat and they are buying back stock, their product will stagnate in 18–24 months unless they acquire. That affects how much you invest in integration vs abstraction.

Hiring and org design: Revenue per employee and operating income trends inform whether your target employer can sustain the team size they are recruiting for. Pair with ground truth from hiring ICs vs managers frameworks when evaluating leadership roles.

Board conversations: When our board asked why we were not matching a competitor's AI feature velocity, I pulled their 10-K R&D trend and capitalized software note. The conversation shifted from "engineering is slow" to "they are spending 19% of revenue on R&D and running negative FCF." Different problem.

Tools

• SEC EDGAR full-text search: company name 10-K
• Diff: diff -u fy2023.txt fy2024.txt or VS Code compare
• Spreadsheet for 3-year trend on 6–8 line items — manual entry, 20 minutes

I do not pay for Bloomberg. For private companies, the equivalent is diligence data rooms and customer reference calls — same mental model, worse formatting.

What I'd do next

Build a vendor watch list with automated 10-K filing alerts (SEC RSS by CIK) and a one-page template the eng leadership fills within 48 hours of a key vendor filing. The template is five bullets: new risks, R&D trend, segment shift, liquidity flag, "does this change our contract renewal stance?"

The goal is not financial analysis. It is early warning for architectural bets that assume a vendor's incentives stay stable.