How are forestry and lumber companies using AI in 2026?

In 2026, forestry and lumber companies are using AI in two places. The first is on machines and imagery: drone footage, harvester cabs, mill equipment. The second is in decisions that cross the business, such as matching production to demand and market prices, which one large producer has now described publicly. What the public record does not yet contain is a measured result for that second kind. Of two of the largest producers' 2026 filings, one lists AI as a lever with no figure of its own and the other does not mention it at all.

What the usual answer says

The top results for this question list applications. A case-study roundup attaches ten big company names to uses from bark beetle detection to demand forecasting, with figures such as "up to 30%" and no source for any of them. A Fastmarkets write-up of a pulp and paper conference in April 2026 reports that "Concrete use cases, like predictive maintenance, process optimization and knowledge capture, are already delivering measurable benefits in yield, energy efficiency and decision‑making speed". It gives no measurement.

The list of uses is roughly right. What it leaves out is how much of it is announced and how much is measured. For a business deciding where to spend, that is the part that matters.

What the companies themselves published

Lesprom, summarising a Wall Street Journal report in April 2026, says Weyerhaeuser "has trained an AI model to review drone footage and measure seedling survival rates". It adds that the company is "using AI to monitor mill equipment, match production with demand and market prices, and optimize truck routes". Lesprom's headline, like others that rank for this question, links AI to "$1 billion in annual profit by 2030".

The company's own documents are narrower. Its June 2026 investor presentation sets a goal of $1.5 billion of incremental Adjusted EBITDA by 2030. "Artificial Intelligence" appears on the growth slide as one lever among many, grouped with integration, automation, cost and supply chain work. Its Investor Day release of December 2025 values that whole group of "enterprise initiatives" at $180 million, with no separate figure for AI.

West Fraser reported its first quarter and second quarter 2026 results in releases of about 4,400 and 4,700 words. Neither mentions artificial intelligence or machine learning. The closest either comes is capital spending on "projects focused on optimization and automation of the manufacturing process".

Why the second kind is harder to prove

AI on a machine has a measure built into its job. A model that counts seedlings or flags a failing bearing can be checked against a count or a repair record. AI that matches production to demand and prices has no single record to check against. Its result only shows once harvest, production, inventory and order data agree well enough to compare a before and an after. Where those sit in separate systems, the comparison has to be built before any result can be seen.

At the Fastmarkets conference, one industry speaker warned against "pilot purgatory". Fastmarkets' report of the talk sums up the condition: "AI only creates impact when it is embedded into everyday operations and linked to clear business outcomes". The public record for 2026 is consistent with that warning. The announcements are there and the outcomes are not yet published.

What a measured claim looks like

A measured claim names the task, the figure before, the figure after and what the result was checked against. From Quarri's own work with a sawmill: a daily production tally that took about a quarter of an hour by hand now takes a couple of minutes, and ties to the dollar against the manager's own sheet. The saving is small. What makes it a result is the tie to a record kept by someone else.

That gives a test for any AI claim in this industry, from a vendor, a peer or a trade article. Ask for the before and after figures, and ask what independent record the result was checked against. A claim that can answer both is evidence, and one that cannot is a plan. By that test, every 2026 claim read for this piece, in filings and in trade press, is a plan.

Where the test doesn't apply

Silence in a results release is weak evidence. Listed companies rarely put unaudited operating metrics in filings, a mill can run machine learning in its scanners without anyone naming it in a quarterly release, and private operators publish nothing at all. This reading covers two listed companies' filings plus the trade press that reports on them, and it is a sample. The test also matters less where the AI's output is itself a direct measurement, such as a log count from an image, which can be checked on the spot.

Quarri for operations teams is built for the people running the crews, the lines and the yard.

Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.