← The Inside Track Kalshi Real money

Will a 1-megawatt data center go live in orbit?

Aired 5 August 2026 · written 1 August, when Before Jan 1, 2035 was 36%

Will a 1-megawatt data center go live in orbit? — episode poster
◆Watch this episode◆
Share on
TikTokXBlueskyFacebook

The market we based this episode on

The board on 1 August 2026, the day the script was written. A price is the market’s odds of that outcome.

Before Jan 1, 203536%
Before Jan 1, 203329%
Before Jan 1, 203122%
Before Jan 1, 203016%
Before Jan 1, 202912%
Before Jan 1, 202810%
Platform
Kalshi · real money
Resolves
1 January 2035
Volume traded
$46,668
Outcomes on the board
6
Lifetime volume on this market is $46,668 — enough to read a trend, but thin enough that a single motivated trader can move the price.
Resolution criteria. the operational deployment in Earth orbit of a data center delivering at least 1 megawatt of electrical power to compute hardware on a single spacecraft or a unified co-orbital cluster operating as one compute system, actively processing workloads

Why we chose it

This bet exists because Nvidia-backed Starcloud launched its Starcloud-1 satellite in November 2025 with an H100 GPU and trained an AI model in orbit for the first time, kicking off an "AI in space" arms race that now includes Google's Project Suncatcher and Jeff Bezos publicly calling gigawatt orbital data centers inevitable within 10-20 years. The 1 MW threshold is the specific chokepoint — it's roughly the smallest scale that counts as a "real" hyperscaler node, not a science demo.

Why the skepticism baked into these odds: Starcloud 3, the dedicated data center spacecraft designed for SpaceX's Starship, is envisioned as a three-ton vehicle with only 200 kilowatts of power — so even Starcloud's flagship near-term ship is 5x short of 1 MW. Starcloud-2, launching later in 2026 with Nvidia's Blackwell B200, has 100x the power of Starcloud-1's ~1 kW class — still ~100 kW, an order of magnitude away.

Bull case for 2030-2033: Starship cadence unlocks 4km-wide solar arrays and Starcloud/Google/Axiom race to plant a flag. Bear case: thermal rejection at MW scale in vacuum is brutally unsolved, and every timeline in space slips.

The show’s read Before making an episode we research the market to judge whether it is worth one. This is what that research said about this bet — our reasoning, not reporting.

What we thought

SalThe Handicapper
EddieThe Philosopher
Sal
A megawatt in vacuum needs a football-field radiator glowing red-hot to shed heat. Starcloud's aiming at a tenth of that. Thirty-eight's rich.
Eddie
You're worried about the radiator, Sal — I'm stuck on the shrug. A third of the room thinks the cloud goes literal by 2035 and just… nods.

How it’s moved since

Daily price. The dotted line is the day the episode was written — the odds Sal and Eddie were looking at. The solid line is the day it aired.

0% 10% 20% 30% 40% written 1 Aug aired 5 Aug 17 Jun 10 Aug 27 Sep Before Jan 1, 2035 38% Before Jan 1, 2033 31% Before Jan 1, 2031 26% Before Jan 1, 2030 20% Before Jan 1, 2029 15%
Written 1 Aug Aired 5 Aug Source: KALSHI

Biggest move since: Before Jan 1, 2031, up from 22% on the day the episode was written to 26% today.