1. Core Financial Questions to Ask About AI Companies
What is the company's current annual recurring revenue from AI products, and how does it compare to its total revenue? For example, Microsoft reported more than $13 billion in annualizedized revenue from its AI cloud services as of early 2024, a key metric investors track closely source. What percentage of revenue comes from AI versus legacy products, and is the mix shifting quickly enough to justify a premium valuation? Ask for the company's gross margin on AI services, because AI infrastructure costs can compress margins if pricing does not keep up with compute expenses.
What is the company's capital expenditure plan for AI training and inference infrastructure over the next three years? Nvidia's data center revenue reached $475 billion in fiscal 2024, reflecting massive hyperscaler spending on GPUs and networking source. How much of that capex is locked into multi-year contracts with cloud providers, and what happens if those contracts roll over at lower rates? Investors should also ask about free cash flow conversion, because many AI companies are still reinvesting heavily and have not yet reached steady-state profitability.
2. Competitive and Technical Questions to Ask
What Differentiates the Company's AI Models or Platforms?
Does the company have proprietary training data, a unique model architecture, or a dominant distribution channel that creates a defensible moat? Tesla's full self-driving dataset, gathered from over 400,000 vehicles, gives it a proprietary edge in autonomous driving AI source. How many parameters does the model have, what is the training compute cost, and how does inference cost per query compare to competitors? Ask whether the company's AI stack is vertically integrated or relies on third-party providers, because dependency on Nvidia or another chipmaker introduces supply risk.
What is the company's current ranking in the relevant AI benchmark, such as MMLU, HumanEval, or industry-specific tests? Rankings shift quickly, so a single snapshot is less useful than a trend over the last 12 months. How many enterprise customers are actively using the product in production, and what is the average contract value and retention rate? A high number of pilots with low conversion to paid deployments signals weak product-market fit despite impressive demo performance.
3. Regulatory, Risk, and Governance Questions to Ask
How Is the Company Managing AI Regulatory Risk?
What specific regulations, such as the EU AI Act or SEC disclosure rules, apply to the company's AI products, and what compliance costs have been quantified? The SEC requires public companies to disclose material risks from AI, including cybersecurity, bias, and intellectual property exposure source. Has the company faced any enforcement actions, lawsuits, or fines related to its AI systems, and what is the total potential liability? Ask about the company's AI governance board, responsible AI team size, and whether model outputs are audited regularly for accuracy and fairness.
What are the company's exposure and contingency plans for adversarial attacks, data poisoning, or model theft? SpaceX and other space-tech firms use AI for autonomous systems where failure carries extreme physical risk, requiring rigorous testing and redundancy source. How much of the company's cybersecurity budget is allocated specifically to AI model protection, and has the company conducted third-party red-team exercises? Investors should also ask about workforce concentration risk, because a small team of key researchers or engineers can control