← All insights

The number that travelled around the industry this year is that Weyerhaeuser will add $1bn of annual profit by 2030 using AI. That isn't what the company said. What they did say is more interesting, and on our own numbers it may be too small.

$1.5bnIncremental Adjusted EBITDA targeted by 2030, against a 2024 baseline. Not profit, and not attributed to AI
$180mThe enterprise initiatives bucket where AI actually sits, as one of five levers, unquantified
16.9%What $1bn represents against Weyerhaeuser's addressable cash cost line
2.8%What it represents per year, spread across the six years to 2030

Company figures are from Weyerhaeuser's FY2025 Form 10-K (filed 13 February 2026), the Q1 2026 Form 10-Q (filed 1 May 2026), and the 8-K exhibits accompanying quarterly results. Target figures are from the investor day release of 11 December 2025 and the investor update 8-K of 25 June 2026. Market data is as at the close on 29 July 2026. Our own rates come from Quarri engagements and are marked throughout as proven or identified.

What Weyerhaeuser actually said

The AI framing comes from a Wall Street Journal report in April 2026. Its wording, which the trade press then carried, was that the company:

plans to use artificial intelligence across logging, replanting, trucking and mill operations as it seeks to add $1 billion in annual profit by 2030 without relying on higher lumber prices.

Wall Street Journal, April 2026

Here is the company's own wording, from the investor day release of 11 December 2025:

The growth plan is expected to deliver $1.5 billion of incremental Adjusted EBITDA by 2030, measured against a 2024 baseline, and includes $1 billion of Adjusted EBITDA uplift from identified growth initiatives.

Weyerhaeuser investor day release, 11 December 2025

Adjusted EBITDA, not profit. Measured against 2024, not today. And it splits by segment rather than by technology: $440m Wood Products, $230m Strategic Land Solutions, $180m enterprise initiatives, $150m Timberlands. The remaining $500m is an assumed improvement in product pricing.

AI appears inside that $180m enterprise bucket as one of five levers, alongside integration excellence, innovation and automation, cost initiatives and supply chain optimisation. It carries no dollar figure on any slide. It then appears a second time under "additional potential catalysts" in the investor update of 25 June 2026, which sits outside the $1bn altogether. Management did not mention AI once on the Q1 2026 earnings call.

One more thing worth noting before the arithmetic. The 2024 baseline was $1,292m of Adjusted EBITDA. In 2025 it fell to $1,021m. So the climb is steeper than the headline implies.

Where this value actually sits

Four pools, in our experience. They are very different sizes, and we have proved some and not others.

Cost2 to 6% · proven

Of the spend you actually examine, not of total spend. Procurement benchmarking on a like-for-like basis. Separately, setup and changeover ran around 45% of one remanufacturing cost pool, with roughly 3 to 6% removable without missing a delivery date.

Margin0.5 to 2.0 points · identified

Of gross margin, from pricing floors judged at account level. Alongside it, business selling below cost ran 1.5 to 5% of revenue, and lapsed accounts belonging to departed reps held peak-year revenue worth roughly half of one year's turnover. The biggest pool, and almost nobody counts it as AI.

LabourReal · hardest to bank

Three days a month back for one finance team, on their own estimate rather than our measurement. A daily production tally going from a quarter of an hour to a couple of minutes. Redeployment rather than headcount, and we have never measured it properly.

ThroughputUnproven · probably largest

We can measure the before state precisely. Around 4 in 10 purchase orders arriving on time across more than 13,000 closed orders. A trim line at 78% downtime. Kiln dwell at a median of nine days across 232 runs. An order book that re-solves in about a second, within 1 to 1.6% of a brute-force optimum. What we cannot yet show is the after state.

Working capital cuts across all four. On one operation around 18% of the stock position turned out to be genuine surplus, and 35% of stocked SKUs had not moved in a year.

Applied to Weyerhaeuser

In FY2025 Weyerhaeuser turned $6,905m of net sales against $5,880m of costs of sales, $545m of selling and administrative expense and $15m of other operating costs. Strip out the $509m of depreciation, depletion and amortisation that no software touches and the addressable cash cost line is about $5.9bn.

Applying our own rates, generously, and assuming roughly half that line is externally purchased goods and services:

For scale, Wood Products runs a 5.7% gross margin, on $4,957m of sales against $4,674m of costs of sales. One point of margin in that segment alone is roughly $50m.

What that is worth as enterprise value

Run-rate EBITDA does not stay on the income statement. It gets multiplied.

Weyerhaeuser's enterprise value at the close on 29 July 2026 was about $22.75bn: a market capitalisation of $17.63bn, plus $5,424m of total debt, less $299m of cash, on the balance sheet as at 31 March 2026. Against trailing twelve-month Adjusted EBITDA of $1,001m, that is 22.7 times.

Timber REITs do not trade on manufacturing multiples, and it is worth understanding why before using one. EBITDA reflects a single year's harvest while the asset is a multi-decade biological stock plus the land underneath it. Deferring harvest in a weak market suppresses EBITDA without impairing the asset, so the multiple inflates in a downturn. Weyerhaeuser has not printed below about 15 times across the last eight quarters. We have used a band of 15 to 20 times, which is conservative against where it currently sits.

$2bn to $9bn
Of enterprise value, from $130m to $450m of run-rate EBITDA at 15 to 20 times. That is roughly 9 to 40% of Weyerhaeuser's entire enterprise value today.

Run the same multiple over the company's own target and the scale becomes clearer still. Delivered in full, $1.5bn of incremental Adjusted EBITDA at 15 to 20 times is worth $22.5bn to $30bn. Weyerhaeuser's entire enterprise value today is $22.75bn.

And what it costs

In FY2025 Weyerhaeuser spent $474m of capital expenditure, including $109m on a single engineered wood products facility. Research and development expense for the same year was $5m.

A data and AI programme at that scale costs single-digit to low-double-digit millions a year, all in: licences, integration, and the people to run it. Against $130 to $450m of annual return that is somewhere between ten and forty times. Against the enterprise value it creates, it is a rounding error on one mill.

The company has reported $92m of operational excellence improvements captured in 2025, and around $210m across 2022 to 2025. Against an implied pace of roughly $167m a year, they are behind, though in the right order of magnitude.

So

The headline was wrong. Weyerhaeuser never attributed $1bn to AI, and $1bn on its own is 16.9% of the cost line, roughly double the best we have seen anyone deliver. Spread across six years it is about 2.8% a year, which is a reasonable question rather than a heroic one.

The more interesting direction is the other one. Our range brackets and then exceeds the entire $180m enterprise bucket, and AI is one of five levers inside it, with the rest booked as an upside catalyst deliberately excluded from the total. A company spending $5m a year on research and development is targeting $1.5bn of incremental EBITDA.

If anything, they are being conservative.

Sources

One caveat worth stating plainly. Our rates are extrapolated from engagements at operations turning tens of millions, applied to a company turning $6.9bn, and percentages do not scale cleanly.

See it on your own data.

Live in two weeks, on the systems you already run.

Request a demo →