OpenAI space war begins? Sam Altman plots to buy stake in rocket company Stoke Space, targeting Musk's SpaceX

👤 45va@Sebastian 📅 2026-04-06 11:22:36

The Wall Street Journal revealed that Sam Altman approached rocket start-up Stoke Space this summer to discuss taking a stake. Although negotiations are currently on hold, OpenAI's huge computing power needs remain a huge challenge.
(Preliminary summary: Sister Mu said "AI is not a bubble": the wealth explosion moment of the Internet is being replicated)
(Background supplement: Google officially launched "Gemini 3"! It has reached the top of the world's smartest AI model, what are the highlights?)

In the ongoing hot AI technology arms race, OpenAI CEO Sam Altman has set his sights on space. According to a report by the Wall Street Journal today (4th), Altman was negotiating with rocket start-up Stoke Space this summer for a controlling stake in order to strengthen its rocket launch capabilities and deploy data centers in orbit.

The idea discussed by many people recently is to use the low temperature of space to reduce cooling costs and support massive AI computing with sufficient solar energy. Stoke Space directly collides with the commercial map of Musk's SpaceX.

Altman has long been interested in the possibility of building data centers in space, arguing that the insatiable demands for computing resources of artificial intelligence systems will eventually require so much energy that the resulting environmental impact will make space a better choice.

However, Dynamic Zone has also reported before. A former NASA engineer and Google Cloud expert explained why establishing a data center in space is a completely unrealistic idea. It is full of challenges in everything from power, heat dissipation to radiation tolerance.

To put it simply: this is an absolutely terrible idea and really makes no sense at all. There are many reasons, but the bottom line is that the electronic equipment required to operate data centers, especially data centers that deploy AI computing power in the form of GPUs and TPUs, are completely unsuitable for operating in space.

Extended reading: Former NASA engineer: Building a space data center is the worst bad idea I have ever heard

Altman’s space idea is on hold

Sources revealed that the current progress of negotiations between Altman and Stoke Space is on hold. Before the technology and supervision are mature, OpenAI will still keep its financial firepower on the earth.

Stoke, founded by former employees of Bezos' Blue Origin, is committed to building a fully reusable rocket and wants to catch up with the goals being pursued by Musk's SpaceX.

Energy bottlenecks under the development of AI

OpenAI has committed to spending $1.4 trillion to expand data centers over the next eight years, an amount that is approximately 70 times its annual revenue of $20 billion. Among them, OpenAI will build five giant campuses with Oracle and SoftBank, with a budget of more than US$300 billion, and develop a new base in Milam County, Texas through SB Energy Global, a subsidiary of SoftBank.

Relevant agreements have been announced in November. Altman bluntly stated in a November Podcast: "Energy is a key constraint on chip manufacturing and AI expansion."

Whether the final data center is located in the Texas wilderness or in low-Earth orbit, the logic behind it is the same: AI models require unprecedented power and cooling. Alphabet's Google and satellite operator Planet Labs have reached an agreement to launch two prototype satellites equipped with Google's artificial intelligence chips in 2027. Whether it can realize the ideal of sending data centers into space remains to be seen.

ملصق:
يشارك:
FB X YT IG
45va@Sebastian

45va@Sebastian

محرر Blockchain والأصول المشفرة، مع التركيز علىتحليلتحليل محتوى المجال والرؤى

تعليق (10)

Yasmin 87منذ أيام
Ein deflationäres Modell ist möglicherweise nicht förderlich für die intraökologische Zirkulation.
Farrah 87منذ أيام
Der Inhalt des Artikels ist gut, unterstützen Sie das Teilen.
Gideon 87منذ أيام
Wie trifft DAO Entscheidungen und stimmt ab?
Helen 87منذ أيام
Die Hype-Komponente überwiegt immer noch die tatsächliche Wertschöpfung.
Sadie 88منذ أيام
Die Einbindung von Vermögenswerten in die Kette ist erst der Anfang, und Ökologie ist die Zukunft.
normannisch 88منذ أيام
In der nächsten Stufe wird die Interoperabilität im Mittelpunkt des Wettbewerbs stehen.
Octavia 88منذ أيام
Künftig wird die Branche stärker auf Effizienz achten.
Gail 94منذ أيام
Der Inhalt des Artikels ist gut und unterstützt die weitere Ausgabe.
Octavia 100منذ أيام
Die Ansichten sind klar und die Analyse ist vorhanden.
Elsa 112منذ أيام
aBFT (Asynchronous Byzantine Fault Tolerance) ist in tatsächlichen Netzwerken schwer zu implementieren.

أضف تعليقا

محتوى شائع