Who Actually Covers AI Companies

Ask any AI company for their media list and you get the same twelve names. TechCrunch, The Verge, Wired, VentureBeat, Bloomberg, The Information, and whichever reporter wrote the last big story they read.

The list isn't wrong so much as flat. It treats AI coverage as a single beat with one set of preferences, when in practice it has split into four groups that overlap in outlet and diverge completely in what they'll write. Pitch the wrong one and you don't get a slow no, you get silence, because what you sent wasn't a story to the person who received it.

Here's the map as it stands in 2026, and what each group is actually looking for.

1. The business and power beat

Bloomberg, the Financial Times, The Information, Reuters, the Wall Street Journal. Bloomberg's Shirin Ghaffary covers the business and policy of the AI ecosystem; Rachel Metz covers the large platforms and the startups pushing into them, having come to the beat from CNN. The FT now runs a dedicated AI editor.

They write about money, market structure, partnerships, regulation, and who is beating whom. Product capability is context, not story.

What reaches them: a funding round with a name-brand investor, a partnership with a company their readers already track, a genuine shift in market position, or a policy fight where you have standing. What doesn't: your model scoring higher on a benchmark.

2. The technical and analytical tier

This is where the real evaluation happens, and it lives increasingly outside traditional publications. Tiernan Ray covers the mechanics and economics of how models are actually built and where they break. Latent Space, Import AI, and The Batch do the same in newsletter form.

They want the thing most companies strip out of their pitch before sending it. Architecture decisions and why. Benchmarks with the caveats attached. What the system does badly. An engineer who can talk without a comms person rewriting the answer.

The trade here is real: this audience is smaller and more technical than a Bloomberg readership, and it is disproportionately made up of the people who decide whether to build on you. For infrastructure companies it is often the highest-conversion coverage available, and it is routinely deprioritized in favor of a mainstream placement that converts worse.

3. The accountability beat

Karen Hao is the clearest example — reporters whose subject is AI's effects and harms rather than its progress. Labor, safety, environmental cost, concentration of power, the gap between what systems are claimed to do and what they do.

They are not a distribution channel for product news, and treating them as one is the most common unforced error in AI PR. They are the right call when you have a genuine position on a contested question and can withstand scrutiny of your own practices. If you can't, don't.

4. Broadcast

The lane almost everyone forgets, and the one with the most different mechanics.

CNBC, Bloomberg Television, and the business networks reach an audience that no technology publication touches — customers, enterprise buyers, public-market investors. But you don't pitch reporters here. You pitch producers, and they are booking a person, not a product.

We know the shape of this because we've built it. For an AI silicon client we assembled five separate broadcast lists — Bloomberg, CNBC, business television more broadly, technology-focused television, and the exchange desks at NASDAQ and NYSE — running to roughly 140 contacts, and almost all of them are producers rather than journalists. That is a completely different contact set from the AI beat, sourced differently and pitched differently.

Broadcast works when you have a recognizable person with an argument. It does not work as a product launch channel.

The newsletter tier is bigger than the publications

This is the part most AI companies get wrong on scale.

The largest AI newsletters are reported to run well past a million subscribers each — The Rundown AI above two million, Superhuman AI around one and a half, TLDR AI around 1.1 million. Those are audiences no technology publication matches. Ben's Bites is far smaller, around 165,000, and holds unusual loyalty because of voice rather than reach.

The catch is redundancy. On a major launch morning the big dailies reportedly overlap around 80% on the lead story. Pitching all of them is one story with four chances at the same slot, and it reads as a blast.

The better structure is one daily for reach and one specialist weekly for depth, chosen by what you actually build. The specialist placement is the one that gets forwarded internally at a company evaluating you.

The list you have is already out of date

The AI beat churns faster than any other in technology, in both directions.

Kyle Wiggers ran TechCrunch's AI coverage until mid-2025, after arriving from VentureBeat. He now leads communications at Ai2. That path — from the most-pitched AI beat in tech to the industry it covered — is common enough now that a media list built a year ago has meaningful decay in it, and the failure is invisible: your email doesn't bounce, it just goes to someone who no longer has that job.

Coverage is also migrating from publications to individuals. Several of the most influential AI writers now publish independently, which means the outlet-level list you licensed doesn't include them at all.

The practical consequence is that AI media lists need rebuilding on a cadence closer to quarterly than annually, and they need to be built around people rather than mastheads.

How to tell which group you're pitching

Before sending anything, answer one question: what would this person's editor call this story?

If the answer is a business development, go to group one. If it's a technical explanation, group two. If it's a societal consequence you can defend, group three. If it's a person worth watching, broadcast. If you can't answer it, the pitch isn't ready, and sending it anyway is how a company burns the small number of AI reporters who matter.

Common questions

Who are the main AI reporters? The beat spans business coverage at Bloomberg, the FT, The Information and Reuters; technical and analytical writers who increasingly publish independently; accountability reporters focused on AI's effects; and broadcast producers at the business networks. No single list covers all four well.

What are the best AI newsletters to pitch? The largest are The Rundown AI, Superhuman AI and TLDR AI, each reported above a million subscribers. Specialist weeklies like Import AI, The Batch and Latent Space are smaller but reach a more technical, higher-intent audience. Because the big dailies overlap heavily on major launches, one daily plus one specialist beats pitching all of them.

How often should an AI media list be rebuilt? Closer to quarterly than annually. Turnover on the AI beat is high, and departures don't announce themselves — pitches to a reporter who has changed jobs simply go unanswered.

Is broadcast worth pursuing for an AI company? When you have a spokesperson with genuine name recognition or a distinctive argument, yes, and it reaches buyers and investors that trade press doesn't. It is booked through producers rather than reporters and rarely works as a product launch channel.

Why do AI pitches get ignored? Most often because the pitch is aimed at the wrong one of these four groups. A benchmark result sent to a business reporter, or a funding announcement sent to a technical writer, isn't a weak story — it isn't a story for that person at all.

Salient PR works with AI infrastructure and AI-native companies including Tenstorrent and Browserbase. Reporter roles and newsletter figures reflect public reporting as of July 2026.