Is the AI Boom Over? Let's Talk About It. — Transcript
Full transcript
- 0:00The AI data center build-out had a
- 0:01terrible night on election night,
- 0:03November 3rd, 2026. AI is not popular.
- 0:07AI is something that most Americans are
- 0:10beginning to feel like
- 0:12is set against the American dream, is
- 0:14set against the idea of individuals
- 0:17achieving what they want to achieve in
- 0:18their lives.
- 0:20AI is being positioned as disempowering.
- 0:23In large measure,
- 0:24Silicon Valley has itself to blame on
- 0:26this. We have had every opportunity to
- 0:29message this differently,
- 0:31and we haven't.
- 0:33We haven't.
- 0:34People are out there mocking us and
- 0:36talking about
- 0:38how this is a victory for ordinary
- 0:42people versus people who want AI to make
- 0:45humans extinct. You're not winning if
- 0:48you're talking about making humans
- 0:49extinct.
- 0:50You're just not. Despite the vote,
- 0:52Americans use AI a lot. There is no sign
- 0:56of disinterest or lack of investment
- 0:58from American companies in AI. It's not
- 1:00that we're going to stop.
- 1:02It's that Americans seem to expect that
- 1:04that AI will come without data centers.
- 1:07They seem to expect that that AI will
- 1:09come [snorts] out of the air. Well, it
- 1:11doesn't.
- 1:12That AI is going to come from somewhere,
- 1:14and that's going to produce some
- 1:14outcomes that look really weird if
- 1:16you're only watching the politics. But
- 1:19this is not a video about data centers
- 1:21per se. This is a video about the second
- 1:23and third-order implications that happen
- 1:26when the underlying compute
- 1:28infrastructure that powers our new
- 1:31economy is constrained.
- 1:35I don't think people voting thought that
- 1:37one through, and that's fine,
- 1:39but I do think we have to think it
- 1:40through and understand what it means to
- 1:43make effective decisions in the next
- 1:44couple of years. A May survey from
- 1:46Edison Research at SSRS found that 65%
- 1:50of American adults had used an AI
- 1:52chatbot in the previous week, and that
- 1:53was 6 months ago, and that was up from
- 1:5652% in February. An October poll from
- 1:59ABC News, Post, and Ipsos found that 72
- 2:03% of Americans opposed a new data center
- 2:06in their community. So, widespread use
- 2:09of AI and widespread opposition to the
- 2:11infrastructure AI depends on can and do
- 2:14exist at the same time in the same
- 2:16population.
- 2:18The election just made the consequences
- 2:19of those two opposing choices very
- 2:22concrete. Communities across the country
- 2:24have voted to block or restrict data
- 2:26center development over and over and
- 2:28over again. In Edgewater, Florida,
- 2:30unofficial results show more than three
- 2:32quarters of voters back to prohibition
- 2:34on new data centers, and that is not
- 2:36unusual. Those are typical margins for
- 2:38what we saw across the country on
- 2:40November 3rd. The industry is going to
- 2:42have to deal with the implications of
- 2:44that.
- 2:45But the customers who need the computing
- 2:47power for AI
- 2:49are still there, and their needs are
- 2:51growing. Google said on its latest
- 2:53earning call that usage through its
- 2:55model APIs is up 40% quarter over
- 2:58quarter and more than four times what it
- 3:00was a year earlier. That's not slowing
- 3:02down. It also said it had started using
- 3:04outside compute capacity because of
- 3:06supply constraints on Google Cloud. That
- 3:08is a super different signal from the one
- 3:10coming out of the ballot box, folks. If
- 3:12we put those two things together, people
- 3:14are using more AI. It's getting harder
- 3:16to build the infrastructure to serve
- 3:17them in the places company expected to
- 3:19build it, and that changes where the
- 3:21dollars are going to go. It changes what
- 3:23becomes valuable in the next couple
- 3:25years and which companies can make their
- 3:27plans work. It also changes the odds of
- 3:30an AI infrastructure bubble. And yes, it
- 3:32gives the people trying to put data
- 3:34centers in space another reason to keep
- 3:36on going. Now look at the companies that
- 3:38have already made commitments based on
- 3:40where they thought the build out was
- 3:42going. This is where the state of Texas
- 3:44in particular gets really interesting.
- 3:46There are specialist cloud companies
- 3:47called Neo Clouds that rent out the
- 3:49powerful chips that are used for AI, GB
- 3:51300s and others. Their business depends
- 3:53on getting those chips installed,
- 3:55powered, and earning revenue. If your
- 3:57plan assumed as a Neo Cloud that you
- 3:59could expand quickly on inexpensive low
- 4:02regulatory Texas sites, a longer or more
- 4:05restrictive approval process may not
- 4:07just change the economics of your plan,
- 4:09it may disrupt it enough that the
- 4:10business isn't viable.
- 4:12Texas was already restricting permits
- 4:14before the election, and the question is
- 4:16now what happens to the duration and
- 4:19conditions of those restrictions after
- 4:22electing a vocally concerned governor
- 4:25around data centers. And for work
- 4:27already underway, what matters is how
- 4:29long do approvals take? How long do
- 4:31power connections take? A building under
- 4:34construction and an operating facility,
- 4:36those are two different things, right?
- 4:37You have to actually get the power on.
- 4:39So imagine you've borrowed money and
- 4:40you've committed to equipment and you've
- 4:42signed a customer who expects capacity
- 4:44by a specific date. A delay pushes out
- 4:47or cancels that revenue. Your financial
- 4:48obligations, they keep arriving. The
- 4:50equipment is still depreciating. You may
- 4:53have to buy capacity from someone else
- 4:55to serve the customer you already
- 4:57signed. That might preserve the
- 4:58relationship, but it's going to eat your
- 4:59margins. The danger in this game is
- 5:02running out of money before you can
- 5:04serve demand the way you expected to.
- 5:06And there can still be plenty of
- 5:08customers who want what you are
- 5:10building, and you will still fail in
- 5:12that position because you can't afford
- 5:14to wait. And this is where consequences
- 5:17start looking almost backward. The
- 5:19company that fails doesn't necessarily
- 5:23take the project with it because AI
- 5:25demand keeps rising. Someone else is
- 5:27going to buy that project. So the buyer
- 5:29that has cash and existing operations
- 5:32can afford to wait longer and snap up
- 5:35those deals.
- 5:36It may be able to serve those customers
- 5:38from another facility while the delayed
- 5:40project gets finished on its own
- 5:41timeline. It can acquire equipment and
- 5:43contracts or a partly completed site at
- 5:45a competitive price the original
- 5:48investors wouldn't have accepted when
- 5:49they thought everything was going
- 5:50according to plan. Meanwhile, this
- 5:52advantage extends anywhere where there
- 5:54are already working data centers. A
- 5:56company with a working data center
- 5:58already has capacity that newer
- 6:00competitors are fighting to bring
- 6:01online. As long as demand is holding up,
- 6:04fewer competing facilities means
- 6:06customers have fewer places to go, and
- 6:08that can support higher utilization,
- 6:10better pricing, lower depreciation
- 6:12rates, and stronger negotiating power
- 6:15for existing data centers. So the
- 6:16backlash against the AI build out may
- 6:19end up, ironically, strengthening big
- 6:21tech. The large operators that have
- 6:23projects exposed to delays
- 6:26can afford them. But the companies that
- 6:27have
- 6:29new investments that are startup like,
- 6:31that are neo clouds,
- 6:33they are going to have a harder time
- 6:35surviving the new construction
- 6:38schedules. In other words, the industry
- 6:40may take a very public political defeat
- 6:42and come out more concentrated around
- 6:44classic big footprint, big company kinds
- 6:47of places. So if you're a smaller AI
- 6:49company and you're trying to compete,
- 6:50you may end up renting more of your
- 6:52capacity from the big players that
- 6:55you're trying to compete with in other
- 6:56spheres. Your access to computing ends
- 7:00up becoming a bigger lever in whether
- 7:03your product is viable, and that lever
- 7:05is something that those big companies
- 7:06are sitting on even if you're competing
- 7:08with them. There's another consequence
- 7:10that points in a different direction,
- 7:11though. Slowing down construction may
- 7:14make the overall boom less likely to
- 7:16collapse from overbuilding. Derek
- 7:19Thompson made this argument after the
- 7:20election, and I think he was right. If
- 7:22demand is growing and politics slows the
- 7:24growth of supply in tokens, demand has
- 7:28more time to catch up, and that reduces
- 7:30the risk that a wave of new facilities
- 7:32arrives before there are enough paying
- 7:34customers to use them. Think about a
- 7:36group of developers all looking at the
- 7:38same growth forecast. Every single one
- 7:40of them decides we're going to go for
- 7:42it. Every individual project is going to
- 7:43look reasonable, but collectively they
- 7:46end up building out way more capacity
- 7:49than the market needs at the moment.
- 7:50That's how you get into a risky position
- 7:53where you have expensive facilities and
- 7:54too little demand. Now, to be very
- 7:56clear, we have not yet seen that, but
- 7:59that was a risk we were worried about
- 8:01from an AI perspective in 2027.
- 8:04And so right now, after November 3rd,
- 8:08we're looking at a world where customers
- 8:09have more time to grow into that
- 8:11capacity, existing facilities can stay
- 8:13busier. So moderating the US build out
- 8:16in 2027 could ironically reduce the
- 8:18chance of a supply glut over the
- 8:20following year or two, even as it
- 8:22creates very, very serious problems for
- 8:25individual developers, especially
- 8:26smaller players. So this can be a
- 8:28situation where a company running out of
- 8:31cash before the opening ends up going to
- 8:34the wall and having to liquidate on the
- 8:36cheap, but net net that effect across
- 8:40the economy ends up moderating the
- 8:42supply of tokens enough that we avoid a
- 8:45bubble situation entirely.
- 8:47Now, I'm not here to tell you what any
- 8:49given stock is worth or what any given
- 8:51Neo cloud is worth. I'm here to help you
- 8:54understand what the implications are of
- 8:57the political choices that Americans
- 8:59made on November 3rd, many of which I'm
- 9:01sure did not cross most voters' minds.
- 9:03You don't have to know what happens in
- 9:042028 to follow the implications here.
- 9:08All you have to do is look at it as a
- 9:10story of demand that has been blocked.
- 9:12Start with a company that needs more
- 9:14compute capacity next year. Literally
- 9:16every company I know. It might be
- 9:17training a model or running an AI
- 9:19service or serving customers who want
- 9:21their agents working all day.
- 9:23If their preferred project gets delayed,
- 9:25that need does not politely wait for the
- 9:28next legislative session. The company
- 9:30looks for another place to get work
- 9:32done.
- 9:33The first alternative might be signing
- 9:35up with a different cloud provider,
- 9:36working indirectly through a different
- 9:38community in the US. Suddenly a location
- 9:40that looked more expensive might be
- 9:42okay. Maybe it has power that can
- 9:44actually be delivered. Maybe the water
- 9:46is there and maybe the community is
- 9:48actually supportive and the developer
- 9:49has a good relationship with the
- 9:50community in the local power station.
- 9:52All of that is valuable in a way that
- 9:55cheap land by itself is not. If you can
- 9:57open on time, you're offering something
- 9:59different than anybody else right now.
- 10:01You're offering a customer of a data
- 10:02center the ability to deliver their
- 10:04product as promised and the cost of
- 10:06missing that date trickles all the way
- 10:09through the supply chain for AI. They
- 10:10think this changes the bargaining
- 10:12position that communities are in. They
- 10:14have something developers increasingly
- 10:16need, a credible place to build.
- 10:19That gives them room to negotiate over
- 10:21things like infrastructure costs, union
- 10:23jobs, local investment, what the
- 10:25developer has to deliver.
- 10:27A developer is going to accept more
- 10:29expensive terms because a dependable
- 10:31agreement is worth way more than more
- 10:34uncertain years trying to get this
- 10:36project off the ground. That makes the
- 10:38usual map of friendly and unfriendly
- 10:40states not super useful. You have to get
- 10:42down to specific sites, look at power
- 10:44agreements, look at local support. If
- 10:47every displaced project piles into the
- 10:49same few welcoming spots on the map,
- 10:52those places are going to face more
- 10:54pressure, too. So ultimately, the
- 10:57advantage is going to go to projects
- 10:59that are already on the ground in
- 11:00friendly states and communities that
- 11:03stayed friendly through November 3rd and
- 11:06that are going to make room for data
- 11:07centers locally. And some of that search
- 11:10because there's just only so much space
- 11:12in the US after November 3rd, that's
- 11:14going to go outside the US. There are
- 11:15countries actively trying to attract
- 11:18that investment. I know that Europe is
- 11:20campaigning for this with their
- 11:22gigafactory initiative. The UK has a new
- 11:24agreement setting out what it expects
- 11:26from data center developers. Net net, I
- 11:29hear much more interest outside the EU
- 11:32from folks developing data centers than
- 11:34inside the EU. And the reason for that
- 11:36is regulatory. Building in the EU
- 11:40is not perceived as being a whole lot
- 11:42faster than building in the US even
- 11:44after November 3rd. And so I think that
- 11:46developers are going to look other
- 11:48places. They're going to look at places
- 11:49that can set their own rules. I think
- 11:51[snorts] Canada is an interesting place.
- 11:54Mexico is an interesting place. They're
- 11:55both geographically proximate to the US.
- 11:58And I think anywhere else in the world
- 12:01that has stable land, stable power, and
- 12:04relatively friendly geopolitical
- 12:06relationship with the US stands to gain
- 12:09from the decisions voters made. So India
- 12:11comes to mind, Southeast Asia
- 12:12potentially, places in South America as
- 12:15well. And jobs will follow that
- 12:16investment pattern wherever it is, in
- 12:18the US or outside it. Someone has to
- 12:20construct the building, someone has to
- 12:22install the electrical equipment,
- 12:23someone has to operate facilities. Some
- 12:25of that employment, some of that
- 12:27spending is going to move abroad while
- 12:30the AI service gets sold back,
- 12:32effectively rented back to American
- 12:34customers. And that's an implication I
- 12:36think a lot of people miss. If you keep
- 12:38a data center out of the community, it
- 12:40doesn't keep the technology out. It just
- 12:42means you're renting the technology and
- 12:44you don't own it. My job is to analyze
- 12:46the second and third order effects that
- 12:47people may not have realized. I think
- 12:48the international shift is a classic
- 12:50example. I doubt anybody was thinking
- 12:52about where data centers go other than
- 12:54not in my neighborhood.
- 12:56But those of us who are involved in AI
- 12:58do have to think about that. Token
- 13:00demand keeps growing. If we're figuring
- 13:03out where to move data centers, we have
- 13:05to start to do that now, assuming that
- 13:08we have an uncertain political
- 13:09environment for at least the next two
- 13:11years in the US and that we have to
- 13:12figure out state by state where we can
- 13:15site projects. And if we can't site them
- 13:17at all in the states, where can we go
- 13:19that isn't the US that's still friendly?
- 13:21This brings me to perhaps the most
- 13:23speculative bet of 2026, which is the
- 13:28increased odds that the orbital data
- 13:30center project is going to get real.
- 13:32Now, we already know that XAI has said
- 13:34publicly they're going after this. Elon
- 13:36has talked about this. Jensen has talked
- 13:38about this. Google already has a project
- 13:40Suncatcher prototype in orbit, which not
- 13:43a lot of people know about.
- 13:45So, there's movement here, but there is
- 13:47a huge difference between a prototype in
- 13:50orbit and whether you can overcome the
- 13:52hardware conditions around radiation,
- 13:55around thermal, around depreciation,
- 13:57around getting and maintenance of your
- 13:58data center to actually get orbital
- 14:00compute live. All of that being said,
- 14:03that is a series of technical
- 14:05challenges.
- 14:07And if you run those down, you are in a
- 14:09much looser regulatory environment in
- 14:12orbit, and that is going to be something
- 14:14that the players who can reach orbit,
- 14:16namely Google and xAI, are very very
- 14:19aware of, and that they are going to be
- 14:21incentivized to lean into after this
- 14:24election. So ironically,
- 14:26this election has, in my mind, increased
- 14:30the odds that we will attack those
- 14:32technical problems with more capital and
- 14:34that we will find out very quickly
- 14:37whether orbital compute works at scale
- 14:40and that we will push on the space tech
- 14:42tree hard enough to get there faster.
- 14:45That doesn't mean that we're going to
- 14:46have orbital compute in two years, but
- 14:49it means that if I were a betting
- 14:50person, my bet would go from 10 years to
- 14:53maybe six to eight because this election
- 14:57creates enough uncertainty around
- 14:59compute in the US that you'd rather have
- 15:01an option in orbit. We'll see. It's not
- 15:04a done deal,
- 15:05but it's something that I think is
- 15:06higher probability now. So if you want
- 15:08to understand what the election means
- 15:10for AI, watch what companies do with the
- 15:12next project they were planning. Are
- 15:14companies waiting? Are they moving? If
- 15:16they're moving, where to? Are they
- 15:17paying more for a location they can
- 15:19deliver on? Are they buying capacity
- 15:21from someone else? Are they investing
- 15:23and producing that capacity somewhere
- 15:25that doesn't have the same constraints?
- 15:27The way companies make those decisions
- 15:29around token supply over the next couple
- 15:31of years will tell you a whole lot more
- 15:34about how committed that company is to
- 15:36AI than whatever they put in a press
- 15:38release.
- 15:40I'm always a fan of reading where people
- 15:42are allocating capital on the balance
- 15:43sheet because I think that tells the
- 15:44truth.
- 15:45And for your own work, be careful about
- 15:48using construction headlines as a
- 15:50forecast for adoption.
- 15:52Your employer is going to keep buying AI
- 15:55services and they will figure out how to
- 15:57get AI from somewhere.
- 15:59The price may go up a little bit,
- 16:01available suppliers may change, but
- 16:03fundamentally the cat is out of the bag
- 16:05on AI and demand is going to keep
- 16:07growing. And so
- 16:10the challenge that I have for all of us
- 16:13is in this world
- 16:15where we very clearly have a lot of
- 16:18political constraints.
- 16:20How do we understand not just the second
- 16:22and third order effects I've discussed
- 16:24here, but understand that this rebuke of
- 16:27data centers came from a place of
- 16:31concern about the future of AI in the
- 16:34American Republic.
- 16:36The United States has been a place where
- 16:38individual freedom is celebrated, where
- 16:40the American dream is celebrated.
- 16:42If AI feels disempowering, if AI feels
- 16:46like it is against the American dream,
- 16:49we're not going to win.
- 16:51It doesn't matter. I had a note after
- 16:53the midterms from a friend of mine who's
- 16:55in tech
- 16:57and
- 16:58he feels left out of the AI race. And
- 17:01get this, he's at an AI company and he
- 17:04doesn't feel caught up. He worries about
- 17:06his job.
- 17:07If that's the world that we're living
- 17:09in,
- 17:10if that's the world we're building, if
- 17:11that's the world we're talking about,
- 17:13we're just doing it wrong. We're just
- 17:15doing it wrong. We need to think about
- 17:17our messaging to
- 17:19voters, to communities across the United
- 17:22States
- 17:24so that we are clear about the benefits,
- 17:26why AI matters, why AI adds value to
- 17:29your day, and doesn't just automate away
- 17:31your job, why AI is not just something
- 17:34you worry about over the Thanksgiving
- 17:36dinner table as an extinction risk,
- 17:37which by the way, we're going to have to
- 17:40talk about at Thanksgiving now. Like
- 17:41that is enough in the public discourse.
- 17:43So we're going to get asked about that
- 17:45all at the same time that people are
- 17:48using Muse more, people are using
- 17:49ChatGPT more, people are using AI at
- 17:51work more. That's the world we're living
- 17:53in. How do we take that demand and make
- 17:56it into demand that people are hopeful
- 17:58about using, bring in American optimism?
- 18:02That's the larger question over the next
- 18:04two years. That's the question we can
- 18:05shape with storytelling. That's the
- 18:07question we can shape with how we talk
- 18:09about empowerment and tooling and
- 18:12training and preparing the American
- 18:14workforce for what's next. And it's a
- 18:15global challenge too. It's not like many
- 18:18in Europe are optimistic on AI. If you
- 18:20look at public surveys in Europe,
- 18:22they're not very optimistic either.
- 18:24We need to think about this globally as
- 18:26an opportunity or we're going to end up
- 18:29fragmenting out what AI could be around
- 18:33a bunch of local challenges to
- 18:37particular AI infrastructure sites. That
- 18:39would be a tragedy. We would be missing
- 18:41the potential of what AI could deliver.
- 18:43And that's what a lot of my channel has
- 18:45been about, the tools I release, Open
- 18:47Brain, Open Engine, talking about some
- 18:50of the work I've done on making my talks
- 18:53and thinking available in the MCP. It's
- 18:55all about basically making it easy for
- 18:58you to get perspective, empower
- 19:01yourself, and get where you want to go
- 19:03and set your goals up. I believe
- 19:04strongly in
- 19:06people being able in this moment
- 19:08especially to use AI to get where they
- 19:11want to go.
- 19:13And I've seen it over and over again
- 19:14that people have been able to figure
- 19:15that out, figure out how to get started,
- 19:18and then they're off to the races.
- 19:20And you can do that too. I'll see you
- 19:22next time. You tell me
- 19:23where you think AI is going after
- 19:26November 3rd, 2026 in the comments.
- 19:28Cheers.
About this transcript
This page contains the full transcript of Is the AI Boom Over? Let's Talk About It. by AI News & Strategy Daily | Nate B Jones, generated from the public captions YouTube serves with the video. The transcript has 3,463 words across 545 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.