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 monthsRecordings per month over the last 12 months — 1 in all, peaking in Aug 2026 with 1.
Appearances
And then the construction cost as well, equally at the end of an S curve, pouring concrete, erecting steel, all of the services around electricity and the plumbing, the cooling of these systems are all quite mature technologies.
And so again, there's question marks there about, well, how do we scale this out, particularly in an environmental context?
where I think there are some fair questions about how much water are we going to require for this aspect?
How much is this going to infringe upon people living around these areas or affect their own electricity cost if we've got this big demand and we can't scale the supply there?
Yes, the maturity and the affordability tends to improve over time, but then peters out as you get incremental innovation tends not to be as good as the initial invention.
If we were to then contrast those inputs with a space-based data center, similar three, but on radically different improvement curves.
So the chips are on a similar one, although we'll have to make some accommodation for radiation hardening of those, which we have experience with in space.
power generation.
There's a lot of solar power available in orbit.
It's there 24 hours a day.
But the solar panels can all be mass manufactured in a factory and bolted onto a satellite.
And so the installation cost and the construction cost of those is quite different to how we think about a solar farm terrestrially.
And then the construction cost is actually a launch cost.
Instead, we don't need to build the physical container and all the gubbins that goes inside it.
We need to put these satellites on an affordable rocket and put them into orbit.
And there is where SpaceX has the real proprietary ability to offer an improving proposition there.
As the Starship rocket gets better and better and more and more scalable, that installation cost for space-based data centers comes down and down and down by orders of magnitude.
And so whilst today it might not be that economically attractive to do AI in space, if you extrapolate the direction of travel both terrestrially and in space for the cost curves, there may very well be a point in the future where it's highly rational to install data centers in orbit.
In fact, it's maybe the more compelling place to do it.
If that is the way we're going, it also means that SpaceX might have a proprietary ability to deliver a greater level of intelligence than other companies can, because the fabric of compute which they have might be inherently more scalable than those of others.
Showing 161–180 of 235 · page 9 of 12
← Previous
Next →