Musk's Orbital Data Center Hype: SoftBank Isn't the Only Skeptic

Elon Musk's ambitious concept of establishing orbital data centers is encountering significant skepticism from key industry figures. At a recent shareholder meeting, Masayoshi Son, founder and CEO of SoftBank, contended that developing space-based data centers would offer minimal cost savings and prove too time-consuming. Son emphasized that “in the battle for AI, the next few years will be far more important than what might happen a decade or so from now,” highlighting the urgency of current technological advancements. This topic, along with OpenAI's strategies for custom chips and chipmaker Groq’s substantial $650 million funding round, was a central point of discussion on a recent episode of TechCrunch’s Equity podcast, featuring Kirsten Korosec, Sean O’Kane, and Anthony Ha.

Kirsten Korosec observed the “very ironic” nature of Son’s skepticism, given SoftBank’s well-documented “long history of wild bets” in various ventures.

Sean O’Kane, on the other hand, suggested that Musk’s proposal to create “a constellation of satellites — satellites that need to be replaced every few years as well — to make up an ‘orbital data center,’” ultimately serves to “guarantee that much more business” for his company, SpaceX.

Delving into the discussion, Sean O’Kane asserted that "neo-clouds are the new oil," signaling a widespread industry shift towards monetizing computing power. He humorously declared, "I’m proud to announce that TechCrunch is now a neo-cloud, give us all your money." O’Kane highlighted that numerous entities are currently "compute constrained," leading any company with the potential to lease out processing capabilities to seize the opportunity. This trend is evident in companies like Groq, which he noted was "semi-hollowed out by Nvidia," and even Allbirds, which reportedly emerged from bankruptcy reinvented as a "neo-cloud provider" rather than a shoe retailer, a transformation detailed in an interview by Tim Fernholz.

This compute-leasing strategy extends to SpaceX, whose initial ambition was to construct an AI platform targeting an addressable market comparable to the U.S. GDP. However, O’Kane noted their interim approach: "before we get there, we’ll just rent out our compute." SpaceX continues this path, having recently secured "their first post IPO deal" to lease compute to a smaller entity, following prior, larger agreements with giants like Google and Anthropic. While O’Kane sees this as a viable short-term business for companies like Groq, he raised a critical question regarding the "durability" of such models in the long run.

Anthony Ha interjected, emphasizing that any discussion of SpaceX's AI and data center ambitions must address Masayoshi Son's recent query: "What is the point of data centers in space?" a question previously raised on the podcast. Ha acknowledged that this sentiment underscores the industry's severe "compute constrained" environment, driving the urgent need for more data centers. Given the Earth-bound challenges in building these facilities, space might appear to be a solution. However, Ha agreed with Son’s "pretty fair points" that even if the concept is technologically feasible—despite "very, very serious" associated costs—it remains a distant prospect, "not happening for years and years and years." Therefore, it offers no immediate remedy for the current demand for data centers.

Kirsten Korosec reiterated SoftBank’s "long history of making wild bets," suggesting it's particularly telling when Son vocalizes a question many others have privately pondered. She noted that while "a lot of VCs and founders have been swept up into the idea of orbital data centers," a concept that might have been dismissed just a few years prior, Son’s high-profile skepticism is a crucial part of the vetting process. Despite its importance, Korosec maintained that it remains "very ironic" that he is the one posing the question, given SoftBank's significant investments in numerous "pretty bold ideas" as reflected in their own pitch decks.

Sean O’Kane chimed in with "WeWork!" as an example of past ambitious ventures, predicting a similar discussion trajectory for orbital data centers in the coming years. He agreed that placing such infrastructure in space presents both "an interesting engineering challenge and certainly an interesting economic challenge." O’Kane acknowledged Anthony Ha’s point regarding Elon Musk's disdain for "red tape" and the absence of "NIMBYs in space," which naturally aligns with his pursuit of such projects. However, O’Kane’s core argument revolved around SpaceX's current business model: its launch operations are "overwhelmingly reliant on Starlink." He contended that SpaceX’s impressive "80 or 90% of the launch market globally" is not solely due to superior services but significantly boosted by Starlink’s demand. Without Starlink, he estimated their market share would likely drop to "maybe 20% or 30%... or 40%, but it certainly wouldn’t be 90%."

O’Kane reiterated his central theme: constructing "a constellation of satellites — satellites that need to be replaced every few years as well — to make up an 'orbital data center'" would effectively "guarantee that much more business" for SpaceX’s launch division, a point he consistently emphasized.

Kirsten Korosec quickly added that SpaceX’s other substantial revenue stream involves "renting out their compute," bringing the conversation full circle back to the broader discussion about chips and computing infrastructure.

Anthony Ha concluded by highlighting a pervasive theme: the concept of "talking your own book." He explained that while not a new phenomenon, executives across all industries, particularly in tech, often articulate future predictions that inherently benefit their own businesses. Ha stressed the importance of remembering this dynamic, especially during discussions about major AI companies amidst "incredible uncertainty" regarding future job markets, environmental impacts, and necessary skill sets.

He noted that while AI CEOs and investors offer insights, their predictions always come with an implied asterisk. Musk's advocacy for orbital data centers, for instance, aligns perfectly with SpaceX's business interests. Conversely, SoftBank is "very, very heavily invested in data center projects here on Earth," influencing Son's perspective. Ha also mentioned Sam Altman, another prominent figure who has "rolled his eyes a bit" at the orbital data center concept, a stance perhaps informed by his "long and complicated history" with Elon Musk. Ultimately, Ha concluded that in this high-stakes environment, there are "no objective, impartial observers here," only individuals with "baggage and tremendous amounts of money at stake."