SXSW puts AI onstage and business models under pressure

SXSW has never been shy about falling in love with the future. This year, the future is showing up wearing an AI badge and will immediately start asking questions about who controls distribution, who owns the audiences, and who’s picking up the utility bill.

SXSW — and the Innovation Conference — runs through March 12–18, and the discussion over AI has wandered even further away from a designated lane and taken over the whole hotel lobby. AI has been here before; SXSW added a dedicated AI track in 2024 and kept it around for 2025. But this year, the framing has changed. The AI lane is now a broader Tech and AI track (sponsored by IBM, of course), and the description says AI and the tech industry have become “inseparable forces.” This year, nearly a third of the PanelPicker proposals touch on AI.

The theme has eaten the conference.

The setting may help. This is SXSW’s 40th year, and for the first time, it has brought its events into the same seven-day window. The Austin Convention Center is out of commission during redevelopment, pushing the festival across downtown into a more scattered, citywide format. But a conference that has always liked cross-pollination now has physically arranged itself so tech people, creators, startup founders, marketers, media executives, and investors will have to trip over one another — on purpose.

Everyone is coming to Austin to talk about AI

A few years ago, AI at SXSW was easy to treat as one more buzzy subject on a long menu of buzzy subjects. Now, AI is showing up as the connective tissue between everything else the festival wants to talk about. Affectiva CEO Rana el Kaliouby is speaking about making AI human-centric. Timnit Gebru (the executive director of DAIR), Karen Hao (the author of “Empire of AI”), and John Palfrey (a law professor specializing in cyberspace) are slated for a section on reclaiming humanity in the age of AI. The keynote from Aza Raskin, the founder of the Center for Humane Technology, asks whether AI can help decode animal communication.

Even before anyone heads to the networking drinks, the program is already pulling the conversation toward power, ethics, cognition, interfaces, imagination — and what exactly humanity is outsourcing to a machine. One session asks what AI agents are doing when no one is watching and frames the problem around work and trust. Another looks at AI in music promotion and the creator economy. Another follows governments — from AI governance talk to actual deployment in services and cybersecurity. A featured session on the real-world impact of AI pulls intellectual-property risk into the same conversation as opportunity. Across the week, AI lands as workflow, distribution, policy, labor, and liability — all at once.

Some of the sessions sound like the demo has already happened and like the harder arguments have arrived. “From Hype to Hard Problems” goes straight at infrastructure, data readiness, trust, and enterprise obstacles. “AI & the Brain” asks what heavy AI use does to thinking. “Habit Hacks” is about getting AI adoption to stick inside organizations. That looks like an argument about public digestion: how tools become systems, budgets, habits, and policy fights.

Search is under pressure. Creative work is under pressure. News is under pressure. Trust is under pressure. The machine isn’t auditioning for attention anymore. It already has it. The week’s sessions keep returning to the same pressure points: agents, recommendation, cognition, implementation, and the question of how much autonomy people are prepared to fork over before convenience starts eating authority. SXSW has always been good at detecting when a tech story is moving from novelty into infrastructure. The software story gets all the applause. The infrastructure story gets the invoice. This year’s AI program says that this year, the invoice is the story.

What they’re really talking about is money

A lot of the shiny AI panels come down to a deeply dull set of unromantic business questions. Who owns discovery once search starts behaving more like an answer machine than a referral engine? Who keeps customer relationships when an AI assistant stands between a brand and a buyer?

The AI agenda looks like a margin fight dressed up in future-tense language. “The Internet After Search” gets at the central nerve: AI agents are changing commerce, information access, and who gets paid for content. That’s a search story, a publishing story, an ad-market story, and a platform-control story — packed into a single session. What comes after search is a distribution story, which is to say it’s a power story, which is usually another way of saying it’s a money story. The anxiety underneath is palpable. If machines can answer queries, complete tasks, and keep users inside their interface (and only their interface), somebody else loses the click, the traffic, the customer relationship — or all three.

The creator economy programming this week at SXSW points to a similar pressure: the idea that communities and content shape commerce. Once AI starts mediating discovery and recommendation, the old scramble for new audience attention could very well turn into an uglier and nastier fight over distribution and platforms. The conversations about AI’s impact on work and creativity often sound philosophical right up until the bill arrives. SXSW’s program circles that tension from different POVs: human-centered AI, humanity in the age of AI, creator-economy programming, questions about cognition, questions about implementation. Everyone wants to know: Will AI make people more valuable, less expensive — or both. Independence has always sounded glamorous in the creator economy. Renting your audience from somebody else’s algorithm sounds a lot less Instagram-friendly.

The conversation is also coming to the physical world. AI infrastructure is expensive and land-grabby. Tech conferences may want to keep the machinery backstage, but SXSW is putting the machinery on the program. The industry is moving beyond the phase where intelligence can be discussed as pure software magic and into the phase where everybody — and we do mean everybody — has to think about land, power, capital allocation, and construction timelines. The machine may be digital, but the spending is brutally physical.

“AI Infrastructure: Building Stargate & Beyond” takes the discussion straight to data center development and the practical question of where this AI expansion can even happen. “The Energy Evolution of Data Centers” asks whether AI can become something more dynamic in relation to the grid. By the time a conference starts putting search economics, data center locating, and electrical load on the same bill as frontier-tech optimism, AI is clearly no longer at the stage where everyone is just passing around chatbot demos and acting impressed.

This year’s conference has very little to do with whether one more chatbot can write one more email. The program keeps pointing at something with harder edges: AI as a battle over distribution, labor, intellectual property, infrastructure, and the right to skim value from every transaction that this software touches. And once the demos end and the lights come up, who gets the revenue — and who gets the invoice?