Building Data Centers in Space

Data Centers in Space: A Discussion with Jim of Phantom Space

Key Concepts:

• Space Data Centers: Establishing data processing and storage infrastructure in Earth orbit.
• Space App Store: A platform enabling the execution of code and data processing directly in space.
• Launch Bottleneck: The limited capacity and control over access to space via launch vehicles.
• Moat (Competitive Advantage): A sustainable competitive advantage, in this case, control of launch capabilities.
• Hyperscalers: Large-scale data center operators (e.g., Amazon, Google, Microsoft).
• Zettabytes: A unit of data storage equal to 10^21 bytes – representing the massive data volumes generated in space.
• Five-Year Replacement Cycle: The projected lifespan of satellites and spacecraft before needing replacement due to software obsolescence.

1. The Vision of a Space Economy & Infrastructure

The discussion centers around the emerging concept of a space economy and the necessary infrastructure to support it. Jim, founder of Phantom Space, envisions a future where space is a new frontier, comparable to the New World 500 years ago. He describes a long-term goal of creating a “space app store” – a platform allowing developers to run code and process data directly in space, similar to how applications function on smartphones. This differs from the approach of companies like SpaceX, which focus on relocating existing large-scale data centers to space, primarily to address power and environmental concerns. He believes this infrastructure build-out is essential for unlocking the potential of space-based data and applications. As Jim states, “over the next hundred years, you know, the space economy will be seen as a frontier, much like the new World was 500 years ago. And this is part of the infrastructure that we have to build out.”

2. Bottlenecks and Challenges: The Launch Problem

The primary bottleneck to realizing this vision is identified as launch capacity. While satellites can theoretically generate kilowatts of power for data centers (compared to the megawatts required for terrestrial data centers), the ability to get these systems into orbit is severely limited. Jim emphasizes that controlling launch capabilities is crucial for scalability and long-term viability. He draws a parallel to historical exploration, stating, “It's kind of like living on one side of a very large river like the Mississippi, and you can't get to the other side without a boat. The rockets are sort of the modern day bridge in our future version of boats that cross the Atlantic or the rivers.” Phantom Space was founded with a built-in launch component, mirroring the original SpaceX playbook of vertically integrating launch vehicle and satellite production. The limited availability of launch sites (requiring access to federal ranges) and the current overcapacity of those ranges are also highlighted.

3. Satellite Lifecycles and Maintenance

The conversation addresses the practical challenges of maintaining infrastructure in space. While chips can technically last 20-30 years, the rapid pace of software development necessitates a five-year replacement cycle, similar to consumer electronics. Jim proposes a strategy of replacing satellites every five years and allowing the older units to burn up in the atmosphere to mitigate space junk. This reinforces the importance of controlling launch to facilitate regular replacements. He notes, “Technically, a lot of these chips can last 20 or 30 years in space. Many of them have. But it's like your iPhone. You know, after a few years, it won't run the software anymore.”

4. Investor Landscape and Credibility

The discussion touches on the investment landscape within the space industry. Jim points out a current phase of “sifting through” companies to identify those with genuine credibility and concrete plans. He highlights the importance of a “very, very high moat” competitive advantage, specifically control of launch capabilities. He notes that investors are increasingly focused on quality teams and business plans, moving beyond the speculative bubble of 2021-2022. Phantom Space benefits from having the “father of the data center concept” on its team, which attracts interest.

5. Hyperscaler Interest and Unique Data Opportunities

Phantom Space is actively engaging with major technology companies ("hyperscalers") regarding potential partnerships. The company has a founder of Campus Data Centers on its board and is in advanced discussions with publicly traded companies for investment and customer contracts. A key driver of hyperscaler interest is the potential to access unique data generated in space, particularly for training large language models. Jim explains, “What we hear from the HYPERSCALERS is they're looking for unique data, right? That they can use to train their large language models and that unique data resides in space.” This data is currently inaccessible due to the challenges of transmitting zettabytes of information back to Earth.

6. Overbuilding and Launch Capacity Concerns

The conversation addresses the risk of overbuilding space infrastructure, mirroring concerns about overcapacity in terrestrial data centers. However, Jim argues that the scarcity of launch capacity will likely prevent a significant oversupply of satellites in the near future. Despite the development of large launch vehicles like SpaceX’s Starship, launch capacity remains a critical constraint. He believes the US government and states like Arizona are actively working to address launch site capacity issues.

7. Funding Trends and the Role of AI

Funding for space projects has become more selective following the asset bubble of 2021-2022. Investors are now prioritizing quality teams and business plans over simply funding rocket development. However, the advent of large language models and their potential applications in space are creating renewed interest and investment opportunities. Jim believes that applying AI to space-based data is a “killer app” and a key driver of future growth. He states, “We think that that will continue to improve and particularly with this advent of. By and large language models and applying it into space.”

Conclusion:

The discussion paints a picture of a nascent but rapidly evolving space economy. While significant challenges remain, particularly regarding launch capacity and infrastructure maintenance, the potential benefits – including access to unique data for AI training and the creation of a “space app store” – are attracting significant interest from both established space companies and major technology players. Controlling launch capabilities is identified as the foundational element for success in this emerging market, creating a substantial competitive advantage for companies like Phantom Space. The future of data processing and storage may well extend beyond Earth’s atmosphere, driven by the demand for unique data and the relentless pursuit of innovation.