Superintelligence investing: evaluate the opportunity beyond the headlines

The US executive order of September 29, 2026 changes executive-branch terminology from AI to Super Intelligence, retaining the existing statutory AI definition for implementation. It does not establish a measured technical breakthrough.

The separately signed White House Accord on Super Intelligence is a voluntary industry commitment concerning safety and review. Signing it is not evidence that an audit has passed, a product has achieved ASI, or a company’s future profits are assured. These are different documents; neither supplies a valuation model.

Start with the business model: chips, cloud capacity, software subscriptions or an application using somebody else’s infrastructure. Ask which customer pays, what could retain that customer, and whether reported sales isolate the relevant activity. A broad cloud segment is not automatically an AI-revenue segment.

Distinguish reported revenue from bookings, backlog, usage metrics and management forecasts. Record the period, currency and definition. If the issuer does not separately disclose the contribution, label the gap unknown rather than assigning all growth to superintelligence.

Capacity expansion can support demand while requiring substantial cash today. Examine operating cash flow, cash capital expenditure, leases, depreciation and funding requirements together. Free-cash-flow definitions can differ; compare like periods and reconcile the measure before ranking businesses.

A historical example: Microsoft’s January 28, 2026 FY26 Q2 call discussed AI investment alongside a year-over-year decline in cloud gross margin, and distinguished cash capital expenditure from finance leases. This illustrates a question to investigate, not a current quote, recommendation or proof of later results.

A capable business can still be expensive. Date the share price and the earnings or cash-flow inputs used in a multiple. Separate trailing results from estimates. A price-to-sales ratio is not comparable across companies with different margins, funding needs and dilution.

Write a base, stronger-demand and weaker-demand case. Vary growth, margins, capital intensity and the valuation multiple independently. Ask what must happen for the current price to be justified and what evidence would contradict that case. These are scenarios, not probabilities or promised returns.

An AI-labelled ETF may overlap with a technology fund, a broad index and individual holdings. Check dated holdings and weights, fees, methodology, currency exposure and any leverage. Several product names can still represent the same underlying companies.

There is no universal safe allocation in this article. The useful question is how much of the portfolio depends on the same customers, suppliers or spending cycle. Direct access to a publicly traded supplier is also different from owning a private model developer. Confirm what the instrument actually owns.

Choose an asset you already follow in AI Investing and ask for a financial overview. Keep the company, period, source, reported result, expectation and missing information separate in your notes. An AI answer can help organise the work; verify material figures in issuer releases and filings, and check timestamps. Coverage varies and data may be delayed.

Save the asset to a watchlist and revisit the thesis after its next reported results. An actionable research note ends with a question and an invalidation condition, not an instruction to trade. This article is educational and does not provide personalised investment advice.

AI Investing is the mobile entry point. If you prefer a desktop workspace for financial research, explore Cala Terminal. For keeping market prices and news visible on a compatible big screen, Market TV serves a monitoring use case. Both are products from A2GROUP.