Can AI Help Investors Price Heat Risk into Company Valuations?

• Heat risk may be mispriced because its financial effects are difficult to isolate.
• AI and machine learning could help investors quantify heat risk by connecting extreme temperatures to company-specific changes in production, costs, margins, and cash flow.
• AI can help investors distinguish firms facing recurring heat-related cash-flow losses from those better prepared to withstand them.

Investors do not need artificial intelligence to tell them that a company operates in a hot location. They need it to estimate what happens to revenue, costs, capital spending, and cash flow when temperatures affect that company’s operations.

That is a much harder problem.

In my earlier article, I argued that markets struggle to price heat because it can reduce an asset’s economic capacity without altering the asset itself. A productive asset, such as a factory or power plant, may remain physically intact while producing less, operating for fewer hours, or costing more to run.

The next question is whether AI can convert that operating loss into a credible valuation input. AI can help, but only if investors have enough company- and asset-level information to estimate how heat affects operations and distinguish that signal from other changes in performance. Its advantage is scale: it can process large amounts of data and identify company-specific relationships that would be difficult to find manually.

The difficulty lies in connecting those relationships to valuation assumptions. A heat-risk score cannot be inserted into a discounted cash-flow model. Investors need to know which financial variables should change, by how much, and through which operating mechanism.

For an investor, the process is to assemble the relevant information, estimate the company’s heat response, and then ask whether that response is already reflected in the market price.

Existing tools already map physical risk at the asset level and convert it into company risk scores. Large language models can extract climate information from corporate reports at scale. The Bank for International Settlements Innovation Hub’s Project Gaia, for example, used them to extract climate-related indicators from 2,328 public documents covering 187 financial institutions between 2018 and 2022, despite differences in terminology and disclosure frameworks. Gaia did not estimate the financial effect of heat. Its relevance here is narrower: it shows how AI can turn scattered climate disclosures into structured data at scale. A different use of machine learning is to search for links between weather and financial performance.

Machine learning is useful because heat damage is unlikely to follow one stable, linear relationship across companies. A small increase in temperature may have little effect until an operating threshold is reached. Beyond that point, heat can reduce productivity and equipment efficiency, increase cooling needs, or constrain production.

The effect can also differ widely across companies facing similar weather. An automated plant with modern cooling and greater operational flexibility may continue operating. A labor-intensive business with older equipment and greater worker exposure to heat may experience a sharp loss of output. Averaging the two can make the overall effect appear modest even when one company faces a material financial loss.

Machine learning can search for these threshold effects and differences across firms more effectively than a model built around one average relationship. A May 2026 working paper by Christian Breitung, Gerard Hoberg, and Sebastian Müller, “Machine Learning the Impact of Climate Change on Firms Worldwide,” illustrates how a machine-learning framework can capture those differences. Their models estimate how abnormal seasonal temperature and precipitation affect sales, efficiency, profitability, and costs, with effects varying widely across firms. The adverse effects are concentrated in more exposed industries, labor-intensive and older firms, and companies operating in less developed regions.

Investors therefore need what could be called a company’s “heat-response function”: an estimate of how production, costs, margins, and cash flow change once temperatures exceed thresholds that matter to its operations. Absent that relationship, investors do not have a valuation input.

That missing company-level estimate matters because it can lead to market mispricing. Breitung, Hoberg, and Müller find that their model-implied weather effects predict returns around earnings announcements. This is consistent with investors failing to incorporate the financial consequences of abnormal weather before companies report their results.

A Management Science study by Carina Cuculiza, Alok Kumar, Wei Xin, and Chendi Zhang reaches a similar conclusion. It finds that firms with greater sensitivity to temperature changes have lower future profitability, appear overpriced, and earn lower subsequent returns as market participants are slow to correct the mispricing. It also finds that sell-side analyst forecasts are less accurate for these firms.

That last result is telling. It is consistent with analysts having difficulty translating temperature sensitivity into earnings forecasts. But these studies, while they link weather sensitivity to firm performance and market pricing, do not give an investor a ready-to-use estimate for a specific company.

One possible explanation for the delayed market response is that heat-related losses often reach the financial statements without a clear label. A hurricane has a name and a date, and its operational effects are usually easier to identify. A hot quarter may instead appear as slightly lower production, higher energy costs, weaker efficiency, or margin pressure. Management may attribute the result to unfavorable mix or operational inefficiency rather than identify heat separately.

Investors may know that unusually high temperatures occurred near a company’s facilities. But that does not reveal whether the heat crossed the thresholds that affect its operations or how large the financial effect may be. The missing information is often the company’s response.

This is where machine learning could add value. It may help estimate the company-specific response before its full effect appears in reported earnings.

Until investors can make such an estimate, the earnings announcement may be where the question is forced. It provides evidence of weaker financial performance, even if it does not isolate heat as the cause. Investors must then decide whether the result reflects a one-off disappointment or a recurring sensitivity that should reduce expected future cash flows.

That distinction has direct valuation consequences. A portfolio manager should therefore ask:

Is the market price consistent with the company’s expected heat-related losses and its ability to reduce them?

A poorly prepared company may be overvalued because the market has not capitalized a recurring operating loss. A resilient company may be undervalued because investors observe its location but do not recognize the measures protecting its operating capacity.

Until analysts can connect changes in operating performance to the heat conditions behind them, recurring heat damage may continue to be classified as temporary noise. The market may begin to recognize the loss around the earnings announcement. The mispricing can persist beyond it.