Pull to refresh
Logo
Daily Brief
Following Why Sign Up
Starcloud raises fresh capital to build AI data centers in orbit

Starcloud raises fresh capital to build AI data centers in orbit

New Capabilities

NVIDIA-backed startup lands a $250 million round to build its largest orbital data-center spacecraft as Earth-based launch slots tighten

August 21st, 2026: $250 million extension, NVIDIA joins

Overview

Updated Aug 22

A dorm-fridge-sized box carrying a single NVIDIA chip ran an AI model in space last year. On August 21, 2026, the startup behind it, Starcloud, said investors handed it $250 million to build something far bigger.

The raise values the Redmond, Washington company at $2.3 billion and pulls chipmaker NVIDIA onto its cap table. The money funds a larger factory and Starcloud-3, its biggest planned orbital data center. The bet: run power-hungry AI in orbit, using constant sunlight for electricity and the cold of space for cooling.

Why it matters

If AI compute moves to orbit, its soaring power and cooling needs shift off Earth's strained grids, but only if launch capacity and economics hold.

Questions about this story

Free account needed to ask — your question is kept and asked for you right after sign-up. Answers are public.

No questions yet — be the first to ask.

Key Indicators

$250M
New funding
Series A extension led by investment firm Manhattan West.
$2.3B
Valuation
Up from $1.1 billion at its March 2026 round.
$450M
Total raised
Cumulative funding since the company's 2024 founding.
1
GPUs flown so far
A single NVIDIA H100 reached orbit aboard Starcloud-1 in November 2025.
88,000
Satellites in the long-term plan
The company's stated goal of roughly 20 gigawatts of orbital compute.

Voices

Curated perspectives — historical figures and your fellow readers.

Ever wondered what historical figures would say about today's headlines?

Sign up to generate historical perspectives on this story.

Log in Sign Up

People Involved

Philip Johnston
Philip Johnston
Co-founder and chief executive of Starcloud
Leading the company's scale-up and its search for launch slots

Organizations Involved

Starcloud
Starcloud
Space technology startup
Raised $250 million to scale orbital data centers

A Redmond, Washington startup building AI data centers that operate in low Earth orbit.
Nvidia Corporation
Nvidia Corporation
Semiconductor and AI computing company
New investor in Starcloud

The chipmaker whose graphics processors power most AI training and now sit at the center of Starcloud's hardware.
Manhattan West
Manhattan West
Investment firm
Lead investor in the $250 million round

The Los Angeles investment firm that led Starcloud's Series A extension.

Timeline

January 2024 August 2026

5 events Latest: August 21st, 2026 · 3 weeks ago
Tap a bar to jump to that date

  1. $250 million extension, NVIDIA joins

    Latest Funding

    Starcloud adds $250 million at a $2.3 billion valuation, led by Manhattan West with NVIDIA and Cisco Investments. Total funding reaches $450 million.

  2. $170 million Series A

    Funding

    Starcloud raises $170 million at a $1.1 billion valuation to expand its orbital plans.

  3. AI models run in space

    Milestone

    Starcloud-1 runs AI inference in orbit and trains a small language model, showing the hardware works in space.

  4. First NVIDIA H100 reaches orbit

    Milestone

    Starcloud-1 launches on a SpaceX rocket carrying a single NVIDIA H100, a computer roughly the size of a small refrigerator.

  5. Starcloud founded

    Origin

    The company is founded in Redmond, Washington, to build AI data centers in low Earth orbit.

Scenarios

1

Starcloud-3 reaches orbit and scales up in-space compute

Possible Resolves by End of 2028

Discussed by: SpaceNews, GeekWire, and TechCrunch coverage of the round

Starcloud books Starship capacity and flies Starcloud-3, its largest spacecraft, carrying far more compute than the single H100 it has flown so far. Success would move the company from single-chip demos to a working orbital data center and validate the solar-plus-cold-space model at scale.

2

Launch bottleneck stalls the plan

Possible Resolves by End of 2028

Discussed by: TechCrunch, citing CEO Philip Johnston on tightening launch capacity

SpaceX's Falcon 9 winds down toward its planned 2028 end while Starship testing slips further. Starcloud cannot secure enough launch capacity on its timeline and publicly delays Starcloud-3. The raise was partly meant to guard against exactly this risk, but money cannot buy rockets that are not flying.

3

A rival flies a multi-chip orbital cluster first

Uncertain Resolves by End of 2028

Discussed by: SpaceNews and DataCenterDynamics coverage of Google's Project Suncatcher and Blue Origin's Project Sunrise

Google's Project Suncatcher, Blue Origin's Project Sunrise, or another entrant puts an operational multi-chip AI computing cluster in orbit before Starcloud-3 flies. That would blunt Starcloud's first-mover claim and shift attention to a larger, better-funded competitor.

4

Starcloud raises again at a higher valuation

Possible Resolves by End of 2027

Discussed by: Venture coverage from GeekWire and Via Satellite

Momentum and NVIDIA's backing draw a larger follow-on round above the current $2.3 billion mark within 18 months. A new raise would signal investor confidence that orbital compute is more than a demo, even before Starcloud-3 flies.

Historical Context

3 moments from history that rhyme with this story — and how they unfolded.

1994-2002

Teledesic's 'Internet in the sky' (1990s)

Backed by Bill Gates and Craig McCaw, Teledesic planned a constellation of hundreds of satellites to beam broadband worldwide. It raised huge sums and generated enormous hype. It never launched an operational network.

Then

The company kept shrinking its satellite count as costs and technical hurdles mounted.

Now

Teledesic suspended construction in 2002 and folded without flying its constellation.

Why this matters now

It shows how a capital-rich, ambitious constellation can collapse under launch and cost realities before reaching orbit, the same pressures Starcloud names as its biggest risk.

1998-2001

Iridium's bankruptcy and revival (1999-2001)

Iridium launched a 66-satellite phone network at a cost of roughly $5 billion. Subscribers never materialized at the scale needed. It filed for bankruptcy in 1999, months after service began.

Then

Investors lost billions, and the satellites nearly de-orbited.

Now

New owners bought the network for about $25 million in 2001 and ran it profitably for years.

Why this matters now

Iridium is a reminder that working orbital hardware can still misjudge demand and pricing, and that a bold constellation can be both a technical success and a financial failure.

2019-present

SpaceX Starlink proves the constellation model (2019-present)

SpaceX began launching Starlink in 2019 and now flies thousands of satellites, using its own cheap, reusable rockets to cut launch costs. It reached millions of subscribers.

Then

Starlink showed large low-orbit constellations could be built and operated at scale.

Now

It reset expectations for what a single company can put in orbit, and made rivals dependent on SpaceX for rides.

Why this matters now

Starlink is both the proof that big constellations can work and the reason launch capacity is scarce, since the same rockets Starcloud needs are booked solid.

Sources

(7)
+1
Business Wire TechCrunch SpaceNews GeekWire DataCenterDynamics NVIDIA Blog TechCrunch