Nodes and Networks:

The Reorganization of American Journalism

by Justin Bank

Introduction: The Reorganization No One Announced

No single event marked the moment American journalism stopped operating as a collection of institutions and started operating as a layered field. It happened incrementally — through layoffs and launches, platform withdrawals and newsletter migrations; through the quiet proliferation of nonprofit wires, civic documentation networks and creator brands built by former institutional journalists who took their audiences with them.

By late 2026, the reorganization is quite legible, if imperfect and incomplete. Between this report’s first draft and final publication, CBS changed hands and its new owner, Paramount Skydance, bought CNN's parent company. Vox Media split into two buyers about a month apart — Vox.com, New York Magazine and the podcast network went to James Murdoch; Penske Media Corp. took the rest, including The Verge and SB Nation in June. Byron Allen closed his own $120 million deal for BuzzFeed and HuffPost a week after the Murdoch sale. And countless independent publishers, on Substack or newsletters or elsewhere, likely came and went without much fanfare.

We set out to study the current landscape of the media business. What we found, as we sorted through these headlines and deal terms, is a field best understood by governance models not by beat, platform or size. By governance, we mean who owns the organization, who makes the decisions and how the money comes in. We identified and named nine cohorts and found that governance turns out to best predict how an organization survives and what it's structurally capable of becoming.

We also discovered a trend linked to these cohorts that we’ll explore in future work about organization size, financial success and certain conditions that tend to lead to sustainability.

Explore the data

I. But first, some economic context

Follow the money. It’s practically the founding scripture of the journalism trade: The idea that if you trace the dollars far enough, you find the truth nobody wanted you to find. It’s a curious, borderline cruel, irony that the industry that coined that phrase has spent 30 years struggling to find its own way.

For most of that time, journalism’s economics were inherited. An outlet was born into a distribution system: a printing press, a broadcast license, a platform’s algorithm. Those operational systems set the foundation for what the business would become. But the ground is shifting faster and faster. That inheritance is long gone. It’s not something you can easily build anymore.

Stated plainly: The size of journalism organizations is trending smaller as publishers go niche to serve more narrowly scoped missions and audiences.

Sort the field by when an organization came into existence, and the picture sharpens: Organizations that predate this era — meaning that they were founded before 1995 when the internet began reshaping print and broadcast distribution — are overwhelmingly operations with over $10 million in revenue. Organizations native to the current era are the mirror opposite. Most sit below $3.5 million in revenue, and only a small share ever crosses $10 million at all.

Those that excel, such as ProPublica, Politico Pro and Semafor, got there through specific types and patterns of capital infusion, not organic growth alone. According to our research, organizations don’t just grow into big companies anymore. They inherit it, or they build a specific machine that targets a specific collection of escalators to get there. And these numbers and patterns probably under-sell the dynamic because there’s a bias for survivorship and no accounting for the countless efforts that failed and folded.

For this report, we pulled together a dataset of 169 organizations, based on their public filings, news or industry coverage and their own announcements. The data is current as of October 2026 and is available on request.

The paper began by tracking these deals, launches and shutdowns as they happened throughout 2026: the Corporation for Public Broadcasting’s board voting to dissolve in January, The Washington Post cutting a third of its staff in February, OpenAI buying the tech talk show TBPN in April, Penske Media absorbing what was left of Vox Media in June, and Kevin Roose and Casey Newton launching Machine Gods with NPR in September. The dataset grew out of the coverage, as it happened, and what that coverage kept showing us, including the patterns among governance types. It is imperfect, in part because of the nature of the changing field, but it gives us confidence in the reorganization we’ve observed.

At this moment, we also have AI tools powerful enough to help make sense of the data. We’ve used Claude to challenge our assumptions and to craft the direction of the narrative — but have used human involvement in every step of the rigorous writing and editing process.

A few examples of the recent shifts:

On Apr. 2, OpenAI announced it was acquiring TBPN — a podcast and YouTube talk show that had been streaming for 18 months — for a reported sum in the low hundreds of millions.
Seven weeks earlier, on Feb. 15, an independent reporter named Roger Sollenberger, publishing alone on Substack, placed the president inside the redacted portions of the Epstein files. Nine days later, NPR had confirmed enough of it that Congress acted and the full FBI case files were released. Sollenberger's own compensation for setting that chain in motion: a GoFundMe that, as of this writing, has raised over $8,800.

Those two numbers sitting next to each other are as good a description of journalism's economics as anything in this paper's dataset.

At the same time, American journalism in 2026 is navigating a multiyear funding collapse that is not one event but many:

Meta ended per-publisher Facebook News tab payments in 2024; referral traffic sits at 32% of its March 2020 peak.
Google’s AI Overviews drove a 33% global decline in search referrals to publishers. Through November 2025, the drop was 60% for small publishers that year.
The U.S. government froze $268 million in U.S. Agency for International Development journalism funding on Jan. 20, 2025.
The Organized Crime and Corruption Reporting Project lost 52% of its budget in a week.
Congress rescinded $1.07 billion in Corporation for Public Broadcasting funding; the CPB board voted to dissolve on Jan. 5, 2026, jeopardizing funding for public broadcasters, which reach 46% of all 213 news desert counties.
The Washington Post lost $177 million over two years and cut roughly 300 newsroom staff members in February 2026.
The Center for Public Integrity closed in March 2025 after 36 years.
More than 136 newspapers closed in 2025 — over two a week, which leaves fewer than 1,000 daily newspapers publishing in the U.S. today. That is the lowest number since researchers at the Medill Center’s Local News Initiative started tracking local news publishers in 2005. In the 21st century, they report that 3,500 weekly and national newspapers have closed.

The pattern is the same across every failure: concentrated dependency. Each organization had over-indexed on a single source — one platform, one funder, one owner — and when that source moved, nothing caught the fall.

The ways in which journalism is constructed, distributed and paid for are changing rapidly. And it is drifting further from the original reporting and fact-finding it had revolved around. A few more examples:

An evolving type of media conglomerate is emerging: Journalism support organizations are funding business capacity rather than editorial output. Traditional philanthropy paid for a reporter or an investigation; the American Journalism Project’s venture-philanthropy model pays for the revenue infrastructure — membership systems, event operations, fundraising staff — that lets a newsroom sustain reporters on its own. The Project’s portfolio generated $160 million in revenue in 2025, a 23% median growth year-over-year.
A new model of civic coverage infrastructure is operating at scale: City Bureau’s Documenters Network paid community members $369,000 in stipends to cover 3,253 public meetings in 2025, at $113 per meeting, across more than 20 sites in 14 states.
Sometimes one journalist breaks out with multiple institutional relationships (and no single revenue line): Casey Newton is in his fifth year publishing the Platformer newsletter, now on Ghost (an open-source blogging and newsletter tool). For three years, he was co-host of the New York Times podcast Hard Fork with Kevin Roose, though Roose announced he was leaving the Times after nine years, telling colleagues he and Newton wanted to “build something we would own.” Hard Fork's final Times episode aired in August. In September, NPR announced it was licensing the resulting show, Machine Gods, so Newton and Roose maintain ownership of the show.
And yet, what’s old is new again: The Associated Press was founded in 1846 as a member-owned cooperative. Today it is publishing personality-driven vertical video on social platforms, signing AI licensing agreements with OpenAI and Google, and developing consumer products, while still running the commercial wire service that has served newsrooms for 180 years.

Thus this paper’s argument: There is no single successor to the integrated newsroom. What has emerged instead are layers — different ownership structures, funding models, production architectures, relationships with audiences. These layers cumulatively do what a single newsroom used to do alone, sometimes by design, more often by accident, and always most efficiently when combination and diversity is an articulated strategy.

Making sense of this requires a map.

II. The cohorts and why a governance taxonomy

It’s worth explaining why this taxonomy is organized by governance rather than by beat, geography, platform or revenue model. The question comes up every time we present this framework to people who know the field, partly because governance means different things to different people. We mean it as shorthand for an organization’s operating structure: who owns it, who makes the decisions and how it’s funded. That amalgam is the variable that allowed us to look at the landscape and see clearer distinctions that explain the patterns we were seeing within cohorts because:

Governance determines so much.

A solo journalist publishing on Substack, Bloomberg sending from their enterprise software and a worker-owned newsroom publishing on Ghost look identical from the outside. They all land in your inbox on a regular basis, generate revenue through their audience’s subscription and could all cover similar beats.

But they are solving different organizational problems. The solo journalist’s ceiling is their own capacity. The co-op’s ceiling is its collective brand-name awareness and group decision-making. One can pivot on a single editorial call; the other requires a vote. One closes when the journalist burns out; the other can outlast any individual member. This juxtaposition of external appearances and operational function led us to examine other modes of categorization.

Revenue model was the first taxonomy we considered, but it quickly became clear that revenue model follows governance, not the reverse. B2B and philanthropic journalism look nothing alike economically, but both tend toward specific governance structures.

Both platform and production model were rejected as taxonomies because they weren’t terribly identifying. Platform tells you almost nothing: The same newsletter platform, like Substack or Beehiiv, can host radically different operations. Production model — how the reporting actually gets made — doesn’t distinguish a nonprofit newsroom from a for-profit one running an identical workflow. Geography as a taxonomy is similar. A worker co-op in New York and one in Denver face the same organizational constraints in different markets.

Governance also predicted something we didn't expect going in: which cohorts could produce organizations with large operating budgets at all, and which structurally couldn't, no matter how long they'd been operating or how good the journalism was. That's not a coincidence of who happened to own what. It's evidence that governance sets real ceilings that show up identically whether the organization is a nonprofit, a co-op or a corporation.

Nine cohorts capture the range of journalism operators currently producing in the United States, in three groups:

Legacy Publishers, Creator Brands and Barbell Hybrids are journalism as most people picture it — reporters, outlets, bylines.
Philanthropic Mesh Networks, Public Media Networks and Collaboration Layers are the infrastructure through which public-minded funding becomes public-interest reporting, and the cohorts most exposed to the funding collapse discussed below.
Worker-Owned Co-ops, Emerging Civic Actors and B2B/Professional Services are the clearest evidence that audiences and the market are finding new ways to fund journalism directly.

Governance: Privately held, publicly traded or family-owned institutional news organizations, plus the freelance economy that feeds them.

What it is. Legacy publishers are the oldest layer of the taxonomy and the one most people picture when they hear “journalism.” They’re defined not by medium but by governance: institutionally owned, editorially staffed, built around the assumption that journalism is a product delivered by an organization rather than a person. What distinguishes the cohort is scale infrastructure — foreign bureaus, legal departments, printing contracts, AI licensing negotiations, pension obligations. The Wall Street Journal and a three-person statehouse nonprofit are both journalism; they are not solving the same problem. The cohort is bifurcating along one fault line: whether it built a direct reader-revenue relationship before platform advertising collapsed.

How it works. The New York Times is the clearest proof of what that bifurcation looks like at the top — $2.82 billion in total revenue (fiscal 2025), built on journalism bundled — The Athletic, Cooking, Games, Wirecutter — and sold to 12.8 million subscribers, 53% on multi-product bundles, alongside a $20 million–$25 million-per-year Amazon AI licensing deal and $13.3 million spent fighting AI litigation the same fiscal year. The bundle captures a subscriber’s entire information budget, and no single-product operation elsewhere in this taxonomy competes for that same dollar.

Ethnic and diaspora media sit in a structurally distinct position — overwhelmingly advertising-dependent, overwhelmingly excluded from AI licensing, bundling and philanthropic capital. A 2025 survey of 72 California ethnic media leaders found 44% earn $100,000–$500,000 annually and 25% under $100,000, even though 50% of Hispanic Americans report getting news from Hispanic outlets at least some of the time.

The freelance economy is the cohort’s invisible, most precarious layer: Rates have been stagnant or declining in real terms since 2010.

How they’re organized.

Primary revenue → subscription or advertising · Secondary revenue → AI licensing, syndication, events, B2B, e-commerce

Advantage → scale, brand trust, bureau access, decades of audience relationship

Vulnerability → advertising dependency, legacy costs, AI overviews (−22% search referrals for large publishers, −60% for small)

Production → beat + assignment desk

Distribution → multi-channel; the larger own it, the smaller borrow it

Named examples: The New York Times ($2.82B total revenue, FY2025; bundle economics at scale) · The Atlantic (profitable ~$100M) · The Washington Post ($177M losses, ~300 staff cut — the platform-dependency case study) · Reuters ($853M FY2024, AI licensing cited as growth driver) · TelevisaUnivision/ViX ($5.1B FY2024; FY2025 revenue fell to roughly $4.8B; largest ethnic-media success case) · Vice/BuzzFeed News (the traffic-dependency failure case: Vice sold for $350M against a $5.7B peak; BuzzFeed sold to Byron Allen for $120M mix of cash plus stock, down from a peak valuation of $1.7B)

Edge case — the Associated Press. Founded in 1846 as a member-owned cooperative, the AP runs 235 bureaus, licenses to OpenAI and Google, and builds consumer products, while roughly 82% of its $500 million–$600 million in revenue still comes from wire licensing. It is the oldest, largest successful journalism cooperative in American history and gets almost no attention in cooperative-model discussions — a mistake this taxonomy names explicitly.

Sub-variant — digital-native institutionalization. Axios sold to Cox for $525 million in 2022 at roughly $100 million revenue. Semafor reached $40 million with events at more than 50% of revenue and 75% margins, valued at $330 million. The Dispatch grew from $5 million toward a projected $10 million in 2025. These aren’t creator brands; they’re institutions built digitally, and their governance looks more like C1 than C2.

The line between Legacy (C1), Philanthropic Mesh (C4) and Public Media (C5) is a gradient, not a wall — States Newsroom content runs in The Washington Post; The Baltimore Banner and The New York Times shared the 2025 Pulitzer for Local Reporting.

Governance: Individual journalists or small teams who own their editorial operation outright, distribute across platforms and earn primarily from direct reader relationships.

What it is. Entrepreneurs building personality-led businesses, like Heather Cox Richardson’s Letters from an American with over 2.9 million subscribers and an estimated revenue of over $5 million. But the cohort is very uneven. Nearly 40% of creator journalists make no money from self-publishing, and only 6% call it full-time income, according to Muck Rack’s “The State of Creator Journalism 2025” survey. This cohort is filled with survivorship-bias. The visible outlets are the ones that survived.

How it works. Isaac Saul built the Tangle newsletter from $850,000 in revenue to $4.15 million with a nonpartisan model producing a 16% conversion rate and 99% retention. But Tangle sits near the top of a steep curve: The median Substack creator earns about $4,000 annually, nearly half earn under $500, and the top 10% capture more than 60% of all subscription revenue on the platform. Platform-level growth tells the same story from the other direction: Substack’s paid-publication count roughly doubled in under a year (50,000 in May 2025 to nearly 100,000 by April 2026), while Beehiiv reported roughly $19 million in 2025 revenue (about $35 million projected for 2026), but Beehiiv’s own first-party data put median free-to-paid conversion at just 0.62% platform-wide. The layer is growing; the median creator inside it isn’t. For every Tangle, there are hundreds who stall, and hundreds more who never reach sustainability.

The defining operational move is poly-platform: Marisa Kabas breaks news on Bluesky, converts that distribution into Beehiiv subscribers and then funds the next investigation with the proceeds. Columbia Journalism Review’s research found a less visible pattern underneath: Many creators treat their platform as a lifeboat rather than a destination, maintained until a better institutional offer appears. Emily Atkin documented 50- to 60-hour weeks; Tara Palmeri called the pace “a treadmill.” The burnout is systematic, and it’s the churn hiding under the survivorship numbers.

How they’re organized.

Primary revenue → subscription/reader revenue, platform revenue share · Secondary revenue → sponsorship, events, licensing, books, speaking

Advantage → zero legacy cost, portable audience, editorial autonomy, speed

Vulnerability → platform-dependent discovery, solo burnout, no legal backstop, income volatility, a ceiling below institutional scale

Distribution → email owned, social for discovery, platform migration as resilience

Named examples. Letters from an American (2.9M+; ~$5M+) · Tangle ($4.15M; 16% conversion, 99% retention) · The Dispatch ($5M→$10M projected; SCOTUSblog acquisition) · The Handbasket (815→4,000 paid in nine months) · Silver Bulletin (~$1.2M–$2M; social 0.7% of traffic) · Extra Points (~$250K–$500K; events as third leg)

Edge case — brand-capital launches. The Bulwark (~90K paid, ~$9M) and The Free Press (~$20M annual recurring revenue before its $150M Paramount acquisition) launched with something most C2 creators lack: institutional brand capital and pre-built audiences. Their trajectories look like creator brands; their launch conditions look like institutional spinoffs.

Sub-variant — the lifeboat creator. Charlie Warzel built Galaxy Brain on Substack, then joined The Atlantic. The platform was a bridge, not a destination — a pattern CJR’s research on “the Substackerati” found underreported.

Edge case — philanthropy finds the boundary. The Lenfest Institute’s News Creator Collaborative — $818,000 across 10 news organizations selected from more than 120 applicants, part of a larger $1.5 million investment, including a Google-Lenfest News Creator Summit Series — funds partnerships between newsrooms and independent creators rather than either alone: Spotlight PA is building creator-led coverage of the 2026 Pennsylvania governor’s race, and a Lexington, Kentucky publisher is training beat reporters to become short-form creators. The dollar figure is small next to the American Journalism Project or Press Forward elsewhere in this paper, but it may be the single most on-thesis funding decision in the field right now — a funder writing a check at exactly the newsroom/creator boundary this taxonomy maps, rather than at either node alone.

Governance: Staff journalists maintaining significant independent platforms alongside institutional employment. The defining question: Who controls the subscriber relationship?

What it is. The barbell is hardest to categorize because it’s often in motion — some resolve toward independence, some back toward institutions, some stay genuinely dual for years.

How it works. Casey Newton has run the most examined dual position in journalism: Platformer on Ghost, a Verge license, a NYT podcast, a jointly built federated experiment (The Forkiverse) — roughly $900,000 in subscription revenue plus a six-figure ad stream. The product argument illuminates what makes this cohort work: Hard Fork exists because Newton is independent enough to bring a distinct voice and institutionally connected enough for the Times to want it in its feed — a barbell product neither side could make alone.

Mehdi Hasan ran in the opposite direction, departing MSNBC for Zeteo (about $3.9 million annualized, over 50,000 paid). Ben Smith left the Times to found Semafor, which has since become a $40 million institution. Puck’s writer-equity model tests whether an institution can retain talent with portable audiences — a test it failed when Tara Palmeri left for her own channel in March 2025, taking her audience with her.

A quieter version of the same resolution is becoming the more common pattern. Kyla Scanlon — a Vanderbilt Policy Accelerator fellow already contributing bylines to The New York Times, Foreign Policy, Market Place and The Wall Street Journal — was formalized as a CNN contributor in June 2026 while keeping her Substack, TikTok and Instagram presence fully intact. It wasn’t a discovery hire; it was institutions absorbing subject matter expert credibility piecemeal — a byline here, an on-air hit there — until a formal title just caught up to a relationship that was already de facto happening across multiple institutions at once.

How they’re organized.

Primary revenue → subscription (independent) + salary (institutional), running simultaneously · Secondary revenue → licensing (often times to a former employer), podcast revenue-share

Advantage → salary as floor, brand capital accelerating independent growth

Vulnerability → employer can change terms; direct audience can atrophy; double-commitment burnout

Named examples: Platformer (~$900K) · Zeteo (~$3.9M annualized) · Semafor ($40M, $330M valuation) · Noosphere (Jane Ferguson, new PBS-adjacent structural attempt) · Puck (~$20M; equity + subscriber bonuses; Air Mail acquisition) · Kyla Scanlon (CNN, June 2026; piecemeal-absorption case rather than a discrete hire)

Edge case — the mainstream talent diaspora. Joy Reid, Jim Acosta, Jennifer Rubin, Don Lemon and Mehdi Hasan all departed major organizations within months of each other in 2023–25, entering the barbell cohort from the top rather than climbing into it — arriving with the brand capital most C2 creators never get.

Sub-variant — Morning Brew. Now fully Axel Springer owned, projecting over $70 million in revenue in 2025, Morning Brew built its model around big-named staff writers whose voices are the product — the barbell dynamic operating inside an institution rather than between one and independence.

Governance: Foundation-funded 501(c)(3) nonprofit newsrooms; editorial control sits with newsroom leadership; revenue comes from foundations, donors and earned income.

What it is. This is the field’s public-interest infrastructure, the layer most explicitly committed to covering what markets won’t fund. The Institute for Nonprofit News now counts 500 member organizations; sector revenue reached $650 million–$700 million in 2024, up 14%. Press Forward has committed $500 million from over 100 funders. Yet the entire sector total equals roughly 1% of newspaper ad-and-circulation revenue from 20 years earlier — the sector has grown while the gap it’s filling has grown faster. A 2026 analysis of 559 funding proposals concluded philanthropic grants “function as life support rather than as capital for transformation.”

How it works. ProPublica’s innovation was financial, not journalistic. It proved impact drives repeat foundation funding. Its most recent IRS Form 990 shows $64.3 million in revenue — roughly 9% of all Institute for Nonprofit News member revenue, up from 7% two years earlier. One organization, and its share is growing.

Texas Tribune is the more replicable counter-case: revenue split nearly evenly across foundations, donors, sponsorships and events, with no source above 25% and an annual festival generating about $2 million from 8,000 attendees that functions as both revenue and editorial product. Its most recent Form 990 shows $15.3 million — proof that diversification alone, done early and evenly, can carry a nonprofit newsroom well past where governance alone would predict. That’s the tension this cohort holds better than any other: Concentrated philanthropic scale and even diversification are opposite theories of resilience, and both work, but they are not the same bet.

The American Journalism Project is testing a third path: venture philanthropy funding business capacity rather than content. Spotlight PA grew from $1.7 million and 12 journalists in 2021 to $6 million-$8 million and more than 20 staff members by fiscal 2024 after AJP investment — the difference between covering the statehouse and covering the statehouse, the budget, the courts and the election at once.

How they’re organized.

Primary revenue → foundation grants, individual donations · Secondary revenue → earned revenue, growing from 18% of sector

Advantage → insulation from advertising, editorial independence, covers what markets won’t

Vulnerability → grant dependence stabilization that doesn’t always enable resilience; donor concentration; winner-take-all dynamics

Named examples: ProPublica ($64.3M; ~9% of all INN revenue) · Texas Tribune ($15.3M; diversified revenue) · States Newsroom ($27.5M as of fiscal year ending December 2024; the fuller trajectory — $23.4M (FY2022) → $17.1M (FY2023) → $27.5M (FY2024) — not a flat number) · American Journalism Project ($160M portfolio; 53 newsrooms) · Center for Public Integrity (closed March 2025 — the concentrated-donor failure case: ~$7.3M (FY2020) → $5.03M (FY2022) → $3.83M (FY2023) → a stub $0.89M (FY2024)) · Spotlight PA ($1.7M→$6M–$8M, fastest documented growth arc post-AJP).

Edge case — journalism schools as producers. UC Berkeley’s Investigative Reporting Program, Arizona State University’s Cronkite News, Northwestern’s Medill News Service and others produce journalism indistinguishable from nonprofit-newsroom output, but it is university-funded and systematically undercounted in sector data.

Governance: Independent stations operated by nonprofit entities, often educational institutions with governing boards. Federally chartered broadcasting, historically funded by the Corporation for Public Broadcasting through federal appropriations; member donations; foundation grants; and underwriting.

What it is. Public media built 50 years of infrastructure no other cohort has: Local editorial products, produced simultaneously nationwide, trusted enough that audiences renew with a credit card every year. The regional collaborative newsrooms built this decade — the Gulf States Newsroom, the New England News Collaborative, the Texas Newsroom — are a genuinely new production model that proved itself just as the funding making it possible was eliminated.

How it works. The CPB board voted to fully dissolve on Jan. 5, 2026. The impact is sharply asymmetric: NPR’s digital revenue grew to over $30 million in fiscal 2025, five times its fiscal 2024 level, with minimal direct federal dependency. NPR will survive. The 467 member stations that depended on CPB in ways NPR never did are a different story: KSOR (Oregon) and WVLS (Virginia) both sit above 90% federal dependency; multiple Alaska stations are near 100%. Public broadcasters reach 46% of all news desert counties and 53% of counties with only one remaining local news source. When those stations go dark, there is nothing underneath them.

How they’re organized.

Primary revenue → member donations, CPB (historically, now eliminated), foundation grants, underwriting · Secondary revenue → digital and podcast revenue (growing), licensing

Advantage → public trust, broadcast infrastructure, editorial independence, owned distribution

Vulnerability → federal dependency varies wildly by station (<1% NPR, 80%–100% rural/tribal); broadcast erosion

Named examples: NPR (~$318M; digital $30M+ FY2025, 5× prior year — can survive CPB dissolution, transformed) · Chicago Public Media ($35.6M per FYE June 2025 990; Sun-Times acquisition, −$12M/yr deficit) · WNYC ($73.9M→$85.3M FYE June 2025, +15.7%, even as it ran a FY2024 deficit and cut 26 staff) · KQED (most-listened U.S. public station; −$8M FY2024 deficit; 57 staffers cut; listenership −26% since 2021) · GBH ($225M campaign launched post-CPB dissolution) · rural/tribal stations (80%–100% CPB-dependent; dissolution is existential, not budgetary)

Edge case — Chicago Public Media. Chicago Public Media is simultaneously a public radio station (WBEZ) and a newspaper owner (Chicago Sun-Times) and recently announced a new community-oriented digital product (Chicago.com). That blur is where some of the field’s most interesting infrastructure is getting built right now.

Governance: Multi-organization collaborations producing journalism jointly, plus infrastructure organizations that fund and coordinate production without producing it.

What it is. The journalism this cohort makes is fluid by design — produced jointly, published under multiple mastheads simultaneously, without a single institutional fingerprint. That fluidity is both its editorial strength and its structural challenge: It reaches further and is harder to fund, sustain or measure than journalism with one owner.

How it works. In 2023, the Missouri Independent, MuckRock and the AP collaborated on radioactive-contamination reporting in St. Louis — no single organization had the public-records expertise, statehouse knowledge and distribution infrastructure alone. ProPublica’s Local Reporting Network operates in 17 states and shared a Pulitzer Prize with the Anchorage Daily News — illustrating how national-local partnerships elevate regional accountability journalism.

These range from high-impact project-based collaborations to permanent operating models. A tension lies in sustainability: What is the survival rate of a journalism collaborative past year two? The Center for Cooperative Media has tracked more than 1,000 U.S. collaborations. Its September 2024 study found 41 permanent ones, which were mostly small, women-led and 100% philanthropy-dependent.

URL Media (over 50 outlets and reaching over 27 million people, profitable four of five years selling collective advertising to Nike, Hulu, McKinsey) and Word In Black pool 40-plus Black and Latino outlets, including Afro and Amsterdam News. BOMESI is now attempting the same model across over 300 Black-owned publishers, backed by a $750,000 Press Forward grant. Marshaling pooled, collective capital-generation has been an approach leveraged, historically, within the Black press in the U.S. It’s not a surprise to see it emerging again with a $45.83 billion addressable multicultural-advertising market emerging rapidly.

How they’re organized.

Primary revenue → philanthropy (nearly all); collectively sold advertising (URL Media, Word In Black, BOMESI — the only exceptions we found) · Secondary revenue → membership dues and per-project sponsorships within tracked networks (Center for Cooperative Media); most permanent collaboratives have no secondary revenue line at all

Advantage → combined expertise and reach no single organization has alone, cost-sharing on investigations otherwise unaffordable solo, distribution multiplied across every participating masthead simultaneously

Vulnerability → Nearly all permanent collaboratives are philanthropy-dependent; no market model exists for coordination infrastructure itself

Named examples: INN/NewsMatch ($400M raised and distributed grants to 500 members) · Report for America ($59.9M raised and distributed; 55% of participants hired permanently) · States Newsroom (39 newsrooms; Creative Commons-distributed) · URL Media (27M+ reached; they report it has been profitable 4 of 5 years) · Center for Cooperative Media tracks 1,000+ collaborations; provides grantmaking, training and infrastructure · ProPublica Local Reporting Network (17 states; 2020 Pulitzer win)

Governance: Editorial staff hold majority ownership and governance rights; revenue comes primarily from subscriptions.

What it is. The co-op is a vehicle for porting a product out of an institution that could no longer sustain it. Defector’s readers were Deadspin’s; 404 Media’s were Motherboard’s; Hell Gate’s were Gothamist’s. In each case, the audience followed a specific cultural sensibility the corporate parent had shown it couldn’t sustain. As Defector’s Jasper Wang has put it, in effect: Worker ownership is a governance model, not a revenue model — revenue still comes from subscriptions, same as before.

How it works. Defector is the most fully documented co-op in American history because it publishes annual financials: Year 1, $3.2 million and 36,000 subscribers (10,000 from Day 1, inherited from Deadspin); Year 3, $4.5 million and 42,102 subscribers; Year 4, $4.6 million and 42,500 subscribers; Year 5, $4.65 million with revenue nearly flat, prompting the launch of limited display advertising. The five-year arc is the clearest data available on what the model produces: rapid scaling off an inherited audience, then a plateau. For now, Defector is the single most fully documented proof that worker ownership is a governance model, not a growth strategy — the audience it inherited from Deadspin set its ceiling on Day 1, and five years of careful publishing hasn't moved it.

404 Media launched on $4,000 in startup capital and was profitable within six months — a figure only possible because the product and its audience already existed at Motherboard. Hell Gate reached profitability in November 2025 (about $850,000 in revenue, over 9,000 paid subscribers, +69% growth in paid subscribers year over year), with co-founder Nick Pinto naming the model’s built-in constraint: “We’re an ice cream stand. We’ve always sold ice cream.” No systematic registry of failed co-ops exists — like the creator dataset, this one reflects survivorship bias.

How they’re organized.

Primary revenue → subscriptions (80–95% of total) · Secondary revenue → podcast/display advertising, events, merchandise

Advantage → aligned incentives, durable predictable revenue, no outside party imposing priorities

Vulnerability → requires a pre-built institutional audience at launch; no incubator comparable to INN; individual departure can be existential threat in small co-ops

Named examples: Defector ($4.65M; only five-year audited public track record in the dataset) · 404 Media (profitable in 6 months on $4,000) · Hell Gate (~$850K; +69% YoY) · Flaming Hydra (60 contributors; explicitly low-ceiling collective)

Edge case — the collective vs. the co-op. Flaming Hydra’s Maria Bustillos names the ceiling outright — “nobody’s gonna get rich” — and treats it as a feature: a publishing community for journalists who don’t need the journalism to be their primary income. This is a legitimate, and entirely different, organizational answer than Defector’s.

Governance: Community organizations, civic intermediaries and documentation networks, usually nonprofit or informal; revenue from philanthropy and civic partnerships.

What it is. This is where journalism and civic infrastructure overlap most completely. City Bureau doesn’t call itself a news organization; El Tímpano doesn’t look like a newspaper; the New Jersey Civic Information Consortium funds journalism without producing any. All of them gather, verify and distribute public-affairs information to communities that markets don’t serve profitably and none has achieved self-sustainability without ongoing philanthropy or civic-partnership funding. That’s not a model failure; it’s an honest reflection of who these communities are.

How it works. El Tímpano organizes coverage around questions arriving by SMS and WhatsApp from Oakland’s Latin American immigrant communities and has generated $600,000 in earned revenue since 2022 from civic partnerships — government agencies and nonprofits paying for trusted community access, inverting the usual funding logic. City Bureau’s Documenters Network pays community members to document public meetings the professional press can’t cover everywhere. The New Jersey Civic Information Consortium — the first publicly funded local-news grantmaking body in U.S. history, created by bipartisan 2018 legislation — has deployed over $12 million across 65 organizations. California replicated it in 2025 with $10 million in public funds plus $10 million from Google, and working groups now exist in Massachusetts, Pennsylvania and Wisconsin.

How they’re organized.

Primary revenue → philanthropy, civic-partnership payments · Secondary revenue → per-assignment stipends, state grantmaking

Advantage → community trust that outside organizations can’t access, lower overhead

Vulnerability → no path to self-sustainability without ongoing outside funding; USAID collapse hit the international/exile sub-variant hardest

Named examples: El Tímpano ($600K earned since 2022) · City Bureau/Documenters Network ($369K in stipends to community members, 3,253 meetings) · Documented NYC (~$3M; four languages) · New Jersey Civic Information Consortium ($12M+ across 65 orgs) · Outlier Media (~$3.1M; via earned civic revenue) · Nowruz Media/AWNA (exile journalism, hardest-hit by USAID collapse)

Sub-variant — state-funded journalism policy. New Jersey’s model is the first systematic state-funded journalism-policy pathway since CPB — bipartisan, survived budget pressure, now replicating in California with active working groups elsewhere. This may be becoming a policy template rather than a New Jersey anomaly.

Governance: For-profit or nonprofit organizations delivering specialized information to institutional subscribers at enterprise, employer-expensed price points.

What it is. This is the structural outlier. The one cohort that consistently breaks revenue ceilings that block every other governance type in this study, which suggests the ceiling is a function of business model, not a law of journalism. The reason: The reader isn’t personally paying. An enterprise seat at Politico Pro is a procurement decision, not a household budgeting choice, and it doesn’t carry the same price sensitivity. Edelman’s 2026 Trust Barometer gives the deeper mechanism a name — an “attachment economy” replacing the old attention economy. Audiences extend trust to a creator not because of reach but because a smaller, values-aligned community feels like their own circle, and that trust measurably transfers. Most people who already trust a financial or lifestyle influencer say they’d extend that trust to something the influencer vouches for. That’s a plausible answer to why employers bother to expense the subscription at all. They aren’t buying information so much as buying access to an audience relationship the institution couldn’t build itself.

How it works. Politico Pro runs at about $100 million on per-seat enterprise pricing for policy professionals. Industry Dive built over $100 million across 26 verticals before its $525 million acquisition by Informa — proof the model scales beyond politics. The newest development is the solo subject-matter-expert creator operating on B2B economics: Ben Thompson’s Stratechery (over $5 million from about 40,000 subscribers at $199–$399/year, widely employer-expensed) and Gergely Orosz’s The Pragmatic Engineer (over $1.5 million from over 1.1 million subscribers) break the ceiling the same way Politico Pro does — the employer pays. Orosz doesn’t identify as a journalist. Whether what he produces is journalism is a genuinely open question; whether it functions as journalism for his audience seems already answered by the market.

How they’re organized.

Primary revenue → enterprise/per-seat subscription, employer-expensed · Secondary revenue → advertising (highest professional CPMs in journalism), events, data/API products

Advantage → consumer price sensitivity mostly absent; lower professional churn

Vulnerability → smaller addressable market; must prove professional utility to justify price

Named examples: Politico Pro (~$100M operating revenue) · Industry Dive ($100M operating revenue at exit; $525M acquisition) · Stratechery ($5M+; B2B pricing logic) · The Pragmatic Engineer ($1.5M+; no journalism training) · The Information (~$15M–$20M; 90%+ subscription) · Morning Brew (B2B division ~$25M of $70M+ projected total)

Sub-variant — Consumer Reports. About $200 million in revenue, nonprofit, ad-free since 1936 — the longest-running, largest proof case for fully reader-funded journalism at scale, and one that gets almost no attention in sustainability discussions precisely because it doesn’t call itself journalism.

The cohorts aren’t nine separate industries. They all include organizations that produce and distribute journalism while reaching audiences and convening communities in various forms. They also can blend between classifications. And we’ve now seen them change over time.

The same disruptions have impacted everyone across the board and led to movement in various directions.

Worker cooperatives in this dataset launched with inherited audiences from a series of institutional failures. Some examples: Defector’s readers were Deadspin’s. 404 Media’s were Motherboard’s. Hell Gate inherited its staff from several mastheads that folded within the same stretch of years, and Aftermath’s from Kotaku.

Bari Weiss went from being a New York Times opinion columnist to writing an independent Substack newsletter called The Free Press to having $20 million in annual recurring revenue to a $150 million Paramount acquisition to becoming editor-in-chief of CBS News – from institutional journalist to independent creator to acquired brand to institutional executive in about five years. Independence, flowing back up into institutional power.

When transitions come in bunches, they usually reflect audience/platform migration as well. Earlier this year, other big names made moves for independence: The Contrarian, Mehdi Hasan’s Zeteo and network stars Joy Reid, Jim Acosta and Don Lemon all nearly simultaneously departed to establish themselves on newsletter and vertical video platforms. It added up to the largest single wave of institutional-to-independent migration in the history of American broadcast journalism — a sentence that would have been unfathomable to consider before the market conditions of this current era made it feasible.

III. Paths to Sustainability

There’s one thing that governance can’t fully explain: Organizations that break past their cohort's typical ceiling almost never do so by getting bigger at the same thing. Across this entire dataset, they seem to cross through one of six specific mechanisms:

A bundle big enough to capture a subscriber's whole information budget (The New York Times, The Wall Street Journal)
An audience captured by ideology rather than reporting (Fox, The Daily Wire)
A newer organization that grew a team and a portfolio at once (Semafor, Morning Brew, States Newsroom)
B2B pricing built for an employer's expense account rather than a household budget (Politico Pro, Stratechery)
Philanthropic scale large enough to function like an endowment (ProPublica, CalMatters)
Genuine diversification, with no single stream large enough to hurt if it disappeared (Texas Tribune, The 19th).

No organization in this dataset sits above $10 million in revenue without having done one of those six things. Getting bigger, on its own, was never one of them. It’s fair to ask whether governance also explains who scales or even survives.

Track the same organizations for three years or more and the pattern holds: Some hold their ground, some cross into real scale for good. Look closely at the middle of the map and the dependence comes into focus. Of the 17 nonprofits between $3.5 million and $10 million in revenue whose filings we could break down, 15 get more than 80% of their revenue from grants and donations. Almost everyone in that range is waiting, to some degree, on someone else's decision.

That’s a story about fragility in one part of the map. The next one is about the map itself.

Line up all nine cohorts and something jumps out that the six mechanisms alone don't explain: Nearly every cohort leans hard toward one end of the revenue distribution.

Legacy publishers sit almost entirely at the top. Worker co-ops and civic newsrooms sit almost entirely at the bottom.

The Philanthropic Mesh, the cohort of nonprofit newsrooms, spreads across the whole range. Every organization in that cohort shares the same governance: mission-driven, foundation-funded, insulated from advertising. And yet the Texas Tribune diversified its way to the top of that range, while nonprofit newsrooms just as old, just as mission-driven, never get close. Same ownership structure. Wildly different outcomes because of scale and monetization strategies. The variables governance can’t account for ARE the whole difference.

That’s the limit of this taxonomy. It’s helpful looking backward. The last 30 years of American journalism treated distribution as a separable problem. First you build the story, then find a pipe for it: a homepage, a search result, a social share, a newsletter, a push alert. Or you relieve the pressures of distribution/monetization being so jointly connected — as many nonprofit newsrooms so often do by making their stories freely accessible to anyone, through the financial support of philanthropists and foundations.

Tom Rosenstiel’s new book, “The Next Journalism,” argues that the era is ending on the production side. Rosenstiel argues that journalism has to stop thinking of a story as a fixed product and start thinking of it as a service built around what a reader actually needs.

The evidence we found in researching this paper says the same fracture is happening on the distribution side, and more violently. Reuters Institute researchers describe it as a shift from “news finders,’’ people who go looking for journalism from a brand they trust, to “news conveyors,” people who encounter journalism secondhand, carried to them by a creator, a group chat or an algorithm. Pew found 84 percent of TikTok’s news consumers get news from people they’ve never heard of, chosen for them by an algorithm neither of them controls.

Governance told us who owns the newsroom, who crosses into real scale, and, mostly, how they’re doing it. It’s telling us less and less about whose economics can sustain them once they’ve arrived or why the capital that lifts an organization sometimes disappears as fast as it came.

Looking forward, it can no longer tell us who owns the 15 seconds of trust between a reader and whomever (or whatever) produced the version of the story that actually reaches them. That’s an emerging gap that will only widen as the internet becomes more intermediated between the parasocial relationships of creators and the protocols of the software being built in this agentic age.

Journalism used to inherently be the platform when it was delivered by a media organization’s own broadcast tower or heaved onto front porches from station wagons sent out from their printing presses. But more and more platforms and intermediaries now sit between a reader/viewer/listener and the journalism that reaches them — from the editors at aggregators like Apple News to the micro-influencers who curate journalism that interest niche audiences. A newsroom’s ability to impact (or even be aware of!) these layers of distribution are uncertain as they continue to evolve so quickly.

There is “no white knight coming,” in S. Mitra Kalita’s words, and there’s no version of this reorganization that arrives at a stable resting place, either. As Kalita bluntly laid out the stakes: “Capitalism relies mightily on democracy. Financial support for the Fourth Estate has never been more important, and it needs to come from all quarters.”

IV. What Comes Next

This paper followed the money to make sense of the current landscape. It found a continually morphing array of nodes and networks. Different arrays of people creating journalism and finding ways to fund it. Governance was the rubric that helped illustrate how we got here. But where we might go next will inevitably be different. Our research leads us to believe it will likely involve continued evolution in the model, with a bias toward lean operations and collaboration as the nodes and networks keep moving and mutating.

Some of that is no longer hypothetical. Cloudflare started letting publishers charge AI crawlers for access in July 2025. The AP, Time, The Atlantic and Reddit were among the first to sign on, and by July 2026, Cloudflare announced its first “pay per use” model that compensates publishers when their work actually shows up inside an AI-generated answer with two pilot LLM companies. There are also a handful of open standards — RSL, Free Our Feeds and the AT Protocol — competing to become the licensing and identity layer the next version of the web gets built on.

And funders are already writing checks at exactly the independent-institution boundary this paper keeps finding: The Lenfest Institute is backing 10 newsrooms this year, chosen from more than 120 applicants, to test creator partnerships rather than pick a side — including a Lexington, Kentucky publisher training its own beat reporters to work as short-form video and audio journalists.

We hope future convenings and scholarship can continue to look into this dynamism with a blend of healthy skepticism and clear-eyed hope. The economics are clearly changing as are the practical realities on the ground where the audience is constantly moving — agentic interfaces already summarizing journalism before a reader sees it, various versions of protocol-and-payments layers being trialed, a fast-growing stack of AI licensing deals, and different visions of community engagement and stewardship.

We believe this environment offers talented reporters, storytellers and operators growing opportunities to figure out what combinations might push news organizations toward new models of sustainability and growth — bundles, B2B pricing, events margins, the audience-owned distribution stack. One thing remains certain: This future will be built by a mix of individuals and institutions working together in new and unimagined ways.

We are all learning this in real time. The field reorganized before anyone announced it. If we keep tracking the dollars that are flowing underneath it, these next shifts will feel less seismic.

V. How we did this work

This report follows the money, but it didn’t set out to.

Through 2026, I tracked the deals, launches, layoffs and shutdowns reshaping American journalism as they happened: the Corporation for Public Broadcasting voting to dissolve, the Washington Post cutting a third of its staff, OpenAI buying TBPN, Penske absorbing Vox Media.

Each event named organizations, and each organization went into a dataset and got its revenue traced: IRS Form 990s for nonprofits, SEC filings and annual reports for public companies, numbers the organizations published themselves, and trade reporting. At first I was looking for patterns in the deals themselves — their shapes and sizes — and for trends by platform or audience. Then I tried revenue models.

None of those explained the field. So I grew the sample, moved the revenue lines, hunted for counterexamples and dropped the weakest estimates. That’s when the pattern around governance emerged, along with the concentration and revenue banding the paper describes.

The result is 169 U.S. organizations — for-profit, nonprofit, public, cooperative and creator-run — sorted into nine governance cohorts and placed on one revenue map. It is a field study, not a census. The data is available on request.

I used AI tools throughout. Claude helped sweep public sources and stress-test arguments. Claude Code ran bulk Form 990 lookups, reconciled versions of the dataset, ran the audits, and built the charts and the explorer directly from the data. People made the judgment calls: defining the cohorts and the six mechanisms, assigning every organization, choosing the source behind every number and deciding what the report says. No figure in the data was generated by a model. I’m responsible for every claim here.

The data has natural limits. Nonprofits disclose far more than private companies and creators do. Governance is associated with where organizations land; the data doesn’t prove it causes the outcome. And as the paper notes, the data kept moving. New financials were filed and reported throughout the months of study. And like we discuss in the study, survivorship bias influences the data too.

Andy Pergam, executive director of the Knight Center for the Future of News, and Carrie Simmons, the center’s communications lead, edited the report and Brian Cleveland copy edited it. Thanks to the operators who publish their own numbers, Defector and Hell Gate among them, and to the builders, funders and journalists who talked these ideas through with me. A disclosure: I co-founded the Independent Journalism Atlas and run Better Media Studio, a consultancy. None of the 169 organizations is a client or a partner of either.

Explore the data

Introduction: The Reorganization No One Announced

I. But first, some economic context

II. The cohorts and why a governance taxonomy

C1 — Legacy Publishers and Freelance Networks

C2 — Poly-Platform Creator Brands

C3 — Barbell Hybrids

C4 — Philanthropic Mesh Networks

C5 — Public Media Networks

C6 — Collaboration and Coordination Layers

C7 — Worker-Owned Co-ops and Collectives

C8 — Emerging Civic and Community Actors

C9 — B2B and Professional Information Services

III. Paths to Sustainability

IV. What Comes Next

V. How we did this work