Luke Ward

speaker

235 appearances 1 recordings 1 series first heard Aug 2026 last heard 4 Aug

Luke Ward’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 1 in all, peaking in Aug 2026 with 1.

Appearances

newest first · ▶ plays the moment

Short Briefings on Long Term Thinking - Baillie Gifford · SpaceX: past, present, future – and finding the next world-changers · 4 Aug 2026 podcast

But that is something which is now practical and plausible for them to do because of the innovations they've made in rocketry.

So if you consider what are the cost inputs to a data center, there's so much attention in the world of finance right now on capital expenditure that's going into compute.

If we assume for the moment that structurally we are going towards a world where the need for compute and intelligence

is a confident thing which we can expect of the future.

And the question is, how are we going to provide that?

And when I look at the traditional and terrestrial way in which we provide for compute, there's three main inputs to that.

There's the compute chips themselves, there's the power which you need for them, and there's the construction costs of building the actual data center itself.

Only one of those three is on a cost curve that has been improving over time.

So Moore's law and the chips, which allows you to double the number of transistors which you're able to fit in a particular area over a particular time period, but also means that the cost comes down at a greater and greater volume.

I think we can be reasonably confident that that is going to continue.

They're keeping advances there.

But when you get to, well, how are we going to power these chips?

A lot of the way in which we make energy as a civilization is at the end of an S-curve rather than the start of an S-curve.

You know, coal power plants aren't getting that much better.

Neither are gas power plants.

Nuclear power plants are quite fixed.

We have had improvements in batteries and solar panels, but again, that's mostly come from installation costs and other sort of ancillary services rather than big breakthroughs in the chemistry there.

All of this is to say that it's hard to see how power can scale at the same rate that compute can scale.

And so that's a big question mark for as we have these projections for building more and more orders of magnitude, more compute.

I don't think anyone's quite got a convincing answer to how we power that all today.

Showing 141–160 of 235 · page 8 of 12