AI is the new heavy industry

**SPEAKER_1** (0:00)
This episode is brought to you by Accenture. When you're advertising operations fall out of sync, everything else follows. Spotify and Accenture are working together to reinvent the rhythm of ad sales. Using automation, analytics, and smarter workflows to simplify campaign delivery and access better data across the business. The result? Less time spent on operations, more time connecting brands with the moments and fandoms that matter most. Learn more at accenture.com/spotify.

**SPEAKER_2** (0:30)
This episode is brought to you by Facebook.
So you were scrolling on Marketplace, and there it was, the bike you'd been searching for. You sent a message and it turned out the seller was super chatty, kind of funny, and an avid cyclist. The next thing you know, you're in a cycling crew. Well, a community cycling group. The thing about Facebook, you might find more than what you're looking for. From a browse to a bike ride, this summer, find more on Facebook.

**SPEAKER_3** (1:01)
This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required compatibility and availability varies 18 plus.

**SPEAKER_4** (1:26)
Roughly 20 million artificial intelligence accelerator chips are currently deployed in data centers worldwide. Right.
And based on current supply chain projections, that number is going to hit 200 million in just a few short years. The deployment rate is doubling every nine months. You are looking at a physical rollout of specialized hardware that has literally no historical equivalent in the technology sector.

**SPEAKER_6** (1:49)
Because we are not talking about consumer electronics.

**SPEAKER_4** (1:52)
Exactly. We are not talking about small devices sitting in your pocket. These are dense, heavy, incredibly complex silicon packages that require massive amounts of supporting infrastructure just to turn on.
Going from 20 million units to 200 million units is a purely industrial problem.

**SPEAKER_6** (2:11)
We really need to stop treating the cloud like it is weightless software. I mean, the terminology we used is a huge disservice to reality. When people hear the cloud, they picture vapor. They picture something ethereal and infinite.

**SPEAKER_4** (2:23)
But the reality is completely different.

**SPEAKER_6** (2:26)
Yeah, this intelligence requires physical land. It requires millions of square feet of poured concrete, dedicated electrical substations, industrial scale water rights for cooling systems, and just an incomprehensible amount of money to function.
Every time you query one of these systems, a physical machine somewhere in a windowless building heats up by a fraction of a degree. It is heavy industry.

**SPEAKER_4** (2:50)
It is heavy industry, and the deployment curve is incredibly steep. We are looking at a tenfold increase in deployed silicon in less than three years.

**SPEAKER_6** (2:59)
To put that in perspective, traditional industrial sectors usually measure tenfold capacity growth over decades. I mean, they plan for gradual expansion.

**SPEAKER_4** (3:07)
Right. The semiconductor fabrication plants that print these chips are currently running at absolute maximum capacity just to meet this curve. They are melting purified sand, slicing it into wafers, and using extreme ultraviolet light to print microscopic city grids onto the silicon 24 hours a day.

**SPEAKER_6** (3:24)
When this hyper accelerated deployment curve collides with the physical world, what actually breaks first in our global supply chain?
If you're not subscribed yet, take a second and hit follow on whatever podcast app you're using. It helps us keep making this. We appreciate you being here.

**SPEAKER_4** (3:39)
Well, that doubling period is expected to fall below nine months shortly.
As silicon becomes abundant, the primary bottleneck shifts. The competitive advantage stops being about just acquiring the processors.

**SPEAKER_6** (3:52)
It comes about what you do with them.

**SPEAKER_4** (3:54)
Exactly. It shifts entirely to architecture. The focus becomes who can extract the most intelligence per watt of power, per dollar of capital expenditure, and per square millimeter of physical space.

**SPEAKER_6** (4:05)
Because you have to optimize the board layout.

**SPEAKER_4** (4:07)
Right. You have to figure out how to arrange the components so the electrical signals travel the shortest possible microscopic distance. At this scale, even the physical distance, a signal travels across a circuit board introduces unacceptable latency and power loss.

**SPEAKER_6** (4:21)
Using a chat bot gives you the illusion of participating in the future seamlessly. You sit on your couch, you type a prompt into your phone, and you get an answer a second later. The interface is completely frictionless.