A few things surfaced in the world of AI this week that I've been contemplating. They showed up in different feeds, different sectors, different conversations. But bear with me - because what looks like three separate stories is actually one, and the ending might be a little more interesting than the headlines suggest.

Americans are souring on AI. The surveys landed this week with a thud - and the numbers deserve a closer look. Pew Research found 52% of Americans are more concerned than excited about AI in their daily lives. Only 10% say they're more excited than concerned. A separate poll found 44% think AI will do more harm than good day-to-day, against 38% who think it'll do more good. Yikes.

But here's the number that actually matters: among Americans who use AI tools at least weekly — which is about one in three adults — a majority see it as a net positive. Among people who rarely or never touch it, the view flips negative... hard. And there's an income fault line running right through the middle too. Among households earning over $200,000 a year, 60% think AI will do more good than harm. Under $50,000? 59% say more harm.

My take is that the story isn't really about whether people have used AI. It's about whether they've seen any of the value land anywhere near them. So far, most of it hasn't.

Which, funny enough, brings me right to the next story.

Right next door to us, in Imperial Valley, a $10–$15 billion, 330-megawatt data center is being pushed through by Imperial Valley Computer Manufacturing LLC. Potentially California's largest. Six million gallons of water daily — in a drought-prone desert community. Energy load that could nearly double the entire county's current electricity consumption. Imperial County approved it as a by-right project, no full environmental review required. The city of Imperial sued. Last month, a federal judge sided with the developer. The project is legally cleared to proceed. The legal fights continue — and as of yesterday, a candidate with ties to the developer just entered the race for the Imperial Irrigation District board.

The economic pitch is real: construction jobs, tax revenue, investment. But here's what doesn't make it into that pitch: most of the value this infrastructure generates flows back to the tech companies that own it. The community absorbs the water draw, the electricity load, the noise, the risk — and watches the returns route back to a campus in Silicon Valley. A friend of mine who works in economic development here in the valley asked me recently: do we need a data center in the valley? It's a fair question. I think the honest answer is: it depends entirely on who gets to keep what it creates. And right now, the track record on that isn't great.

There's a term for this. Digital colonialism — infrastructure arrives, value leaves. And the numbers bear it out. There are 188 organized opposition groups across 17 states. An estimated $98 billion in data center projects were blocked or delayed in a single quarter of last year alone. Twenty-five projects were cancelled outright in 2025 and roughly 99 active projects are currently facing sustained community pushback.

So here's where it gets interesting — and where I think the valley should be paying attention.

Some of you have already heard of OpenClaw. Quick version: an Austrian developer named Peter Steinberger built a personal AI agent as a weekend project in late 2025. It went viral in January, hit 200,000 GitHub stars, and by mid-February Sam Altman had hired Steinberger to lead the next generation of personal agents at OpenAI. The project stayed open source. The premise is simple: AI that runs locally on your own hardware, connects to the tools you already use, and keeps your data on your machine. No cloud required.

This week, NVIDIA is running hands-on sessions at their GTC conference in San Jose where attendees can set up their own local AI agent — on a laptop or their personal NVIDIA DGX Spark device — and walk out with something running. When companies the size of OpenAI and NVIDIA start building lanes for personal, locally-hosted AI, that's not a niche developer story anymore. That's a possible guide-post.

Here's what ties all these stories together and why I think it matters for us here in the 760.

The communities most resistant to AI aren't resisting the technology — they're resisting the extraction. The value goes up, the cost stays local. That's the real trust gap, and no PR campaign or slick sales pitches to the nine city councils is going to close it.

The more interesting question for the Coachella Valley isn't whether we need a massive data center that comes with short-term economic benefit and leaves with a long-term bill. It's what changes when individuals, small businesses, and schools here have access to their own on-prem, local AI — running on hardware they own, with data that doesn't leave the building. The intelligence stays in the community. The value stays in the valley.

Whether the nine cities in this valley will work together to make sure the benefits of AI actually stay here — with residents, families, and small businesses — is the most important local question nobody is asking yet.

Sat Singh hosts SunshineFM daily from Rancho Mirage. He uses AI. He wants the Coachella Valley to own it, not rent it from Silicon Valley. Most probably an idealist.