AI’s real challenge for publishers isn’t extinction; it’s speed and agility

When I first met Alex Kantrowitz, he was still on the media side covering the tech giants that govern our work and our lives from inside BuzzFeed. Then, in April 2020, he released a book: Always Day One: How the Tech Titans Plan to Stay on Top Forever. Of course, this was right as the world shut down. So, everything he’d lined up—launch events, big press hits, the whole rollout shebang—evaptorated.

That kind of moment forces a decision. And Alex made his. No more depending on someone else’s platform. No publisher, no newsroom, no PR gatekeepers. It was time to build his own thing. Now he runs an independent media business, Big Technology, a newsletter and a podcast focused on what the big tech players are really doing, and what it means for the rest of us.

At Base.Camp Red Red Rock Canyon earlier this year, I sat down with Alex for an exclusive conversation on what’s actually happening inside the companies that are redefining how attention is captured and monetized.

Forget Legacy, Start Over… Then Do It Again

When Alex Kantrowitz sat down with Mark Zuckerberg, he expected the usual CEO theater, including 25 minutes of pre-cleared talking points and a PR person watching for raised eyebrows. Instead, Zuckerberg flipped the script. He handed over a draft of an internal strategy doc and asked for feedback.

That moment of Zuckerberg opening the door for input from outside the building set the tone for what would become Alex’s book, Always Day One. The idea, borrowed from Jeff Bezos, is simple: act like your company is just getting started. Don’t defend the past. Don’t protect the legacy. Build like none of it matters.

It’s not just talk. At Facebook, Zuckerberg was hiring young product managers and actively trying to understand what mattered to the next generation of users by actually using the same apps himself.

“He was always asking for feedback,” Alex told me. “Snapchatting with PMs, testing dating apps, trying to understand behavior directly.”

That culture of reinvention isn’t optional anymore.

As Alex put it:

This is where publishers get stuck. We talk about transformation, but we do it like it’s a department, not a mandate. You can’t treat reinvention like a project. It has to be your default mode. The companies leading this next wave aren’t doing it for fun. They’re doing it because survival demands it. If your plan is to coast on what worked before, the clock’s already run out.

AI Investment Is Focused on Infrastructure, Not Consumer Tools

Ask most people where the AI arms race is happening and they’ll point to the apps. The chatbots, the copilots, the endless parade of tools promising to automate your inbox or write your blog posts. But Alex says that’s the surface layer. The real fight is deeper. It’s in infrastructure.

This year alone, the biggest technology companies will spend over $300 billion in capital expenditures. That number doesn’t just reflect ambition. It reflects a bet. “Not all of that is AI,” Alex told me, “but it’s a huge amount. And the reason they’re doing it is simple: if they get this right, the upside is infinite.”

That’s not entirely hyperbole either. Venture capital is literally modeling this moment as a zero-or-infinity bet. Either these investments change the game entirely, or they flame out. But even if the math is based on wild uncertainty, the scale of the spending tells you everything. These companies believe they’re building the next operating system for the economy.

And while consumer-facing AI products are multiplying (each with their own voice, personality, or gimmick), the long-term value isn’t in the front-end. It’s in what powers it. Hosting, compute, APIs, custom chips. The companies controlling that layer will get paid no matter which bot you talk to.

“This is the fiercest, most interesting tech battle I’ve seen in a decade,” Alex said. And he’s not wrong. But the problem is, most of the media and publishing world is still focused on what these tools can do for them, not on what’s actually being built underneath. That’s a dangerous blind spot.

If the money’s moving to infrastructure, and you’re still thinking in terms of content widgets, you’re playing the wrong game.

The Value Exchange for Archived Content Is Over

Ask someone in publishing how they feel about AI, and you’ll hear some mix of fear, frustration, and legal strategy. The idea that these models were trained on the open web (on years of content we created, curated, and fought to monetize) feels like theft to some, and like the end of the value chain to most.

But as Alex told me, the companies building these tools aren’t thinking about it that way. “They believe they have a right to it,” he said. “Because they’re transforming it. The model is literally called a ‘Transformer.’” That mindset is everywhere. The engineers working on these models talk about “downloading the internet” the way you’d describe syncing a playlist. It’s casual. It’s already done.

Alex noted:

It’s not that the archive wasn’t valuable. It was. But value only matters when there’s leverage. And in this new environment, the leverage is gone. “The idea that your historical content holds leverage? That’s done. They’ve got it now. It doesn’t matter if we support it or not. It’s theirs.”

That doesn’t mean publishers are out of moves. But we need to stop acting like the original content alone still holds leverage. The companies building these tools already moved forward. And if we want a say in what happens next, it won’t come from looking backward.

Urgency and Voice Are Still Publisher Advantages

AI can generate almost anything... well, except what it hasn’t seen.

That’s where publishers still have room to move, according to Alex. “I don’t think you can replace urgency with AI,” he told me. “If you’re publishing something the model hasn’t seen yet, you’ve still got an edge.”

In a world where large language models (LLMs) are trained on what’s already out there, real-time news becomes one of the last defensible formats. If your newsroom is built to move, you’re not competing with the archive. You’re delivering something AI can’t replicate.

Because it hasn’t happened yet.

The second advantage is less operational and more personal. It’s voice. “The web’s been commoditized for a long time,” Alex said. “Everything’s designed to rank or optimize. But AI’s going to flatten that even more.”

What he means is: if your content still reads like it was built for Google in 2015, you’re about to get wiped out. But if you’ve got a distinct point of view, and if readers know there’s a person behind the words, you’ve still got a lane.

No one’s asking ChatGPT for Bill Simmons’ opinion. They’re asking it to summarize what he said. That difference matters. We’ve spent years automating personality out of publishing in the name of scale. The next phase is going to reward anyone who can bring it back. If your content has a voice people trust (or even just recognize), you’re ahead of the machines for now.

Publishers Can Start Turning Content Into Tools

The idea that content and software are merging isn’t some abstract future scenario. It’s already happening. And right now, publishers are sitting on formats that could evolve into real, usable products.

Alex has seen it firsthand. While prepping for an interview with Meta’s CTO, he didn’t have time to finish reading five articles. So he dumped them into an AI tool that turned them into a podcast, custom, automated, and designed entirely for him. “That’s not a podcast a publisher would ever make,” he said, “but it was exactly what I needed.”

It’s a small example, but it shows where things are headed. AI tools let users generate personalized experiences that traditional content teams would never build because there wasn’t enough scale to justify it. That’s the long tail, and it has the potential to explode.

On a flight, Alex pushed the idea even further. He uploaded a fake bank statement and prompted the model to build a financial plan, adding just two words: “use code.” What came back was surprisingly powerful. “It gave me a bar chart, analyzed the spending, and even built a retirement calculator based on my inputs,” he said. “It was better than any financial calculator I’ve found online.”

This wasn’t prebuilt software. It was live, contextual, created from content, according to Alex:

This is where publishers need to shift their mindset.

Your content alone isn’t the product anymore, it becomes your starting point of how you define and provide value. Value, in this context, comes from what your audience can do with the information you’re giving them. If you can offer tools that let them engage, test, calculate, or customize in real-time, you’re no longer fighting for clicks.

You’ve created a utility. Utility is what earns repeat behavior. Utility is what makes you a part of someone’s workflow, not just a part of their feed.

Yes, AI Is a Pressure Test for Publishers … but It Doesn’t Have to Be an Extinction Event

There’s a lot of anxiety in publishing right now. And fair enough. Audiences are shifting, platforms are unreliable, organic exposure is nonexistent. Now, AI is cutting into the middle of it all. But what’s coming isn’t a mass wipeout.

It is, however, a reckoning. The question is whether your operation can keep up when the cycle speeds up and the expectations shift.

The companies Alex talks to (those that are shaping the next wave) aren’t cutting teams just because they can. They’re pushing harder because AI gives them the capacity to do more. “We’ve been resource constrained,” he said, quoting one exec. “There’s so much more we want to do, but we haven’t had the bandwidth.”

This is what publishers like you need to hear:

AI is going to expose every place you’re slow. Every outdated workflow. Every approval loop that kills a story’s shelf life. If your system can’t handle a daily news cycle, a data-driven story, or a custom tool, someone else will ship it first.

As I said during my conversation with Alex, if you’re still clinging to your old cycles, your old stack, or your old way of thinking about content, you’re already behind. You don’t have to beat OpenAI. But you do need to beat your own inertia.

AI is the filter. If you can’t adapt, it’s going to become obvious. But if you can move faster, test smarter, and rethink how you deliver value, there’s still plenty of ground left to claim.