How is Louisiana-Pacific using technology?

Through its products, its measurement of its mills and, so far only lightly, AI. First, product technology: engineered wood siding and OSB, with expertise in "wood composites, overlays, chemical treatments" and paints. Siding made up 62% of 2025 net sales of $2,708 million. Second, manufacturing measurement. LP publishes overall equipment effectiveness for each business, measured at every mill against a common target. Siding ran at 77% OEE in each of 2023 to 2025, and OSB rose from 75% to 79%. In the first half of 2026, siding reached 84%. Third, AI, which LP says "remains in the early phases" in its operations, in wording other companies also use. Of the three, the measurement is the most telling.

What the usual answer says

Descriptions of LP's technology cover its engineered products, advanced manufacturing, automation and digital tools for quality and efficiency. The products are well documented. In the 10-K, the word automation appears only in the auditor's section on IT controls. The filing says much more about how LP measures its mills.

A published operating metric

LP's 2025 Form 10-K, filed on 17 February 2026, includes an unusual section. LP measures the OEE of each mill "to track improvements in the utilization and productivity of our manufacturing assets". OEE is defined as "a composite metric that considers asset uptime (adjusted for capital project downtime and similar events), production rates, and finished product quality". LP uses "a best-in-class target across all LP sites that allows us to optimize capital investments, focus maintenance and reliability improvements, and improve overall equipment efficiency".

The results are published by business. Siding was 77% in 2023, 2024 and 2025. OSB was 75% in 2023, 78% in 2024 and 79% in 2025. The 2024 10-K also showed a South American line, dropped in 2025 when the segments changed.

The Q2 2026 10-Q continues the table. Siding OEE was 85% in the second quarter against 83% a year earlier, and 84% for the first half against 81%. OSB was 79% for the first half against 78%. Half-year figures aren't directly comparable with full years, since the second quarter of 2025, at 83%, sat inside a year that averaged 77%. The first-half comparison is like for like, and it is up.

LP cautions that "other companies may present OEE data differently". In our reading, a single best-in-class target across sites only works if OEE is defined the same way at each of them.

What LP says about AI

The AI disclosure is brief and cautious. "Our development, integration and use of AI technology in our operations remains in the early phases." An EDGAR full-text search finds the same sentence in other companies' 10-Ks, so it is standard risk-factor language rather than a measured self-assessment. It also appears in LP's 2024 10-K. The risk factor then lists what could go wrong. There could be new cyber vulnerabilities, leaks of confidential information through AI tools, and intellectual property risks. It also names "potential legal or reputational harms due to insufficient or flawed data, inaccurate or misleading outputs, insufficient quality control, or unlawful bias". The phrase "inaccurate or misleading outputs" is new in 2025. The filing adds that AI tools "may not generate the intended efficiencies".

The Q2 2026 10-Q says nothing further on AI.

Why the metric matters more than the label

OEE is a summary: uptime, rate and quality combined into one ratio per mill. It isn't what a downtime or maintenance model learns from. Those models need event-level records, such as stoppage logs, sensor readings and work orders. What carries over from OEE is the discipline behind it. A ratio can only be compared across sites if every site records downtime, rate and quality the same way. If those underlying records are defined differently, any model built on them learns the differences in definition.

LP's own AI risk factor makes the same point from the other side. Flawed data and misleading outputs are among the harms it names. A company that already measures its mills consistently is better placed to avoid them.

The same logic applies at a smaller scale. 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 tie to a record someone else keeps is what makes operating data trusted.

What it means for other manufacturers

LP's disclosures describe one metric, defined once and reviewed against one target across every site. Behind any such metric sit the event-level records that AI would actually use. LP's filings show the first. They say nothing yet about the second.

What to watch

Siding OEE has now risen in 2026 after three flat years. The filings don't say why. If LP later credits analytics or AI for any of it, the OEE table gives a baseline to judge that claim against.

When it doesn't apply

Filings describe what is disclosed, and automation and digital tools may be in use without mention. Publishing OEE is also an investor-relations choice: it shows measurement, not how clean the records beneath it are. OEE also suits continuous manufacturing better than batch or custom work.

Quarri for wood products is built around how a wood products plant runs, where the order book meets real capacity.

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.