14 publishers collected $15.08B in APCs as annual open-access spend quadrupled

Estimating global article processing charges paid to 14 publishers for open access between 2019 and 2025

Lisa Matthias, Eric Schares, Juan Pablo Alperin, Leigh-Ann Butler, Sherry Kuang, Nina Schönfelder, Stefanie Haustein

cs.DL

2026-08-17

List prices from 14 publishers times OpenAlex OA counts yield $15.08 billion in APCs for 2019–2025; inflation-adjusted annual spend rose from $0.93 billion to $3.74 billion.

What problem this solves

Open access moved the paywall, then put the bill on authors and libraries. Fully OA (gold) journals charge an article processing charge for every paper. Subscription journals sell OA one article at a time (hybrid). How much actually changed hands is not public.

The most precise ledger, OpenAPC, still covers only 482 institutions and 287,105 paid invoices through 2026. DOAJ skips hybrid journals and has no historical fees. Earlier estimates often applied today's list price to older volumes, which inflates the past. This paper lines up annual list prices from 14 large publishers and multiplies them by OpenAlex counts of APC-eligible OA articles for 2019–2025.

Method

The price file covers ACS, CUP, De Gruyter, EDP, Elsevier, Frontiers, IEEE, IOP, MDPI, OUP, PLOS, Sage, Springer Nature, and Wiley: 12,540 journals and 69,856 journal-years. Eleven publishers are complete for all seven years; De Gruyter and OUP start in 2021; IOP misses 2020. Frontiers, MDPI, and PLOS are gold-only.

Non-flat fees are handled conservatively. Required add-ons are included; page charges assume a 10-page article; voluntary and overlength surcharges are dropped. Country-scaled discounts and EDP's pay-what-you-want Liberty APC are excluded, because a typical payment cannot be recovered.

Gold journals treat filtered research and review articles as APC-eligible and ignore OpenAlex's OA tag, which is known to flicker. Hybrid journals count only articles labelled hybrid OA. That yields 4.21 million gold articles and 1.38 million hybrid articles. Spend is list price times count, then converted to 2025 dollars with the IMF Advanced Economies CPI. Both nominal and real figures are reported. Publisher-year article counts in OpenAlex sat within 2% of Dimensions when the authors last checked.

Every article is assumed to have paid the list price. Discounts, waivers, and transformative agreements are not subtracted. The authors' defence: those agreements are usually prepaid from library budgets on volume or legacy subscription, so the community still pays.

Results

Nominal spend over seven years is $15.08 billion. In 2025 dollars it is $16.22 billion. Annual real spend rose from $927 million in 2019 to $3.74 billion in 2025, a fourfold increase.

YearGold (2025 USD, millions)Hybrid (2025 USD, millions)
2019684243
20221,673702
20252,3941,343

Gold still takes 66.7% of real spend, but hybrid grew faster: gold +250%, hybrid +451%, with hybrid article volume up 427%. That runs against OA policy, which spent the same years trying to leave hybrid behind. Transformative agreements are named as a likely driver. Springer Nature says its agreements produced ten times more OA articles in 2024 than author choice alone.

In 2025, 87% of the money went to five houses: Elsevier, Springer Nature, MDPI, Wiley, and Frontiers. That share sat between 85% and 87% every year since 2019. Elsevier grew 638% in real terms, from third place to first, and crossed $1 billion in 2025. MDPI briefly overtook Springer Nature in 2021, then sat at $620–630 million from 2022 to 2024 while output fell 17.4%; higher per-article fees held revenue flat. Frontiers dropped 25.4% from 2022 to 2023, lining up with paper-mill scandals and Web of Science delistings. Wiley closed 19 compromised Hindawi titles and still showed no dip in gold revenue.

Market-wide spend grew 303% against 238% article growth, so fee inflation did part of the work. In 2025 the median listed gold APC was $2,160 versus $3,550 hybrid. Weighted by articles those medians become $2,900 and $3,810: output concentrates in dearer journals. MDPI's listed median is $1,450, paid $3,140; Frontiers $2,350 to $3,800. ACS and PLOS go the other way.

Why it matters

This is one of the widest public APC ledgers, and it prices each year with that year's list rather than back-casting today's fees. Libraries, consortia, and funders get a number they can recompute. Dataset and code are open.

The policy signal is sharper than the dollar total. Hybrid is growing faster than gold. Spend is growing faster than papers. The APC is still tracking prestige and market power more than production cost.

It remains an estimate of list-price exposure, not publisher cash received. Treat the billions as a floor on the conversation, not an invoice.

Limitations

The authors list the hard gaps. List price is not net receipt; author-level waivers are invisible. Taylor & Francis has no bulk price list, so one of the Big 5 is missing and both global spend and concentration are understated. Transformative agreements are not modelled separately, which overstates what individual authors paid even if the community total still holds.

OpenAlex hybrid tags sometimes hit near 100% OA. Manual checks of those titles found genuine hybrid rather than delayed OA, but there is no published error band. Gold journals count all research and review items, including true zero-fee titles; 1,989 journal-years are listed at $0.

Two authors disclose a financial interest in pending litigation against four of the publishers in the sample.

Terms

Source

What people are saying

Related papers

All paper explainers