Can “rocket-grade” engineering capabilities reduce dimensions and strike? Musk says: SpaceX (SPCX.US) has an advantage that Google and other companies cannot reach on the AI circuit

The Zhitong Finance App notes that in the first quarterly earnings report issued by SpaceX (SPCX.US) as a listed company, what is particularly remarkable is its huge investment in building a new type of AI computing power. The company's AI-related capital expenditure doubled from the previous quarter to $15.8 billion, and management said it expects to maintain a similar level of spending by the end of this year.

Admittedly, compared to the investment of hyperscale cloud giants such as Amazon (AMZN.US), Alphabet (GOOGL.US), and Microsoft (MSFT.US) in data centers, this capital expenditure is still dwarfed. In the last quarter alone, each of these three companies spent between $41 billion and $55 billion. But they also generate huge revenue from their respective cloud computing divisions, and have a large number of signed and unfulfilled businesses.

However, SpaceX CEO Musk believes that thanks to a unique competitive advantage, his company can deploy capital more effectively than those cloud giants.

SpaceX's strengths in AI

Musk argues that in the cloud computing field, SpaceX has an engineering advantage over everyone else. His premise is that the company can use the engineering talent pool and intellectual property accumulation that supports the development of its rocket business to efficiently build a new type of data center that can generate a high return on investment.

Musk said in the company's first performance conference call, “We have found that even applying a very small portion of the experience we learned in building rockets (which is extremely difficult) to data centers can generate huge benefits,” he specifically pointed out that the company's cooling system far exceeds current needs.

Significant return potential

Chief Financial Officer Brett Johnson said in a prepared statement: “The current economic benefits have translated into our ability to achieve a payback period of less than a year in deploying new computing power capital.”

This is in stark contrast to Amazon CEO Andy Jassi's statement, which outlined the economic benefits of the large-scale expansion of Amazon Cloud Services (AWS).

He explained, “The capital investment in the data center began two years before we were able to put in servers and start monetizing.” This is just a physical limitation faced in building a new data center; it can't even start to monetize within two, let alone balance the balance of payments within a year.

“For servers and networking equipment, it takes an average of a little less than three years to balance these investments,” Jassi added. Even if the construction cost of the data center was zero, Jia Xi said it would take almost three years to balance the balance of payments just by purchasing the equipment equipped with these buildings. The results announced by Microsoft and Alphabet also indicate that their operations have similar timelines.

There is a huge gap between SpaceX's accounting processing and Amazon's accounting processing. Is a company like SpaceX, which only stepped into the cloud computing field almost yesterday, actually far more efficient in terms of engineering and capital deployment, to the extent that it can generate three times the performance of cloud giants?

Investors are skeptical about Musk's claims

Musk has never shied away from making bold assertions about the direction of his business development and when it has reached various milestones. But the actual results often fell short of his predictions. We have reason to doubt that with engineering talent alone, SpaceX will be able to establish a substantial competitive advantage in the field of AI computing power.

While Johnson's statement about the current extremely fast payback period may be accurate, it's unclear whether it can scale up just as efficiently. SpaceX has existing infrastructure, which was used last quarter to provide more computing power to third party customers such as Anthropic and Alphabet's Google. This may not necessarily be replicable.

More importantly, in the face of competition from cloud giants, it makes logical sense that SpaceX can easily retain such top talents. As previously mentioned, Alphabet, Amazon, and Microsoft are investing three times more in capital expenditure than SpaceX. Their business relies heavily on an efficient return on that capital expenditure. If they can increase the return on capital several times, they are bound to pay high salaries to top talent.

SpaceX was only in a position to provide some computing power at a time when there was a serious shortage of AI computing power. Alphabet signed a computing power contract with the company because the long-term potential of using its own infrastructure to immediately serve large customers like Anthropic by transferring some of its internal AI computing power requirements to a third party is too good to give up.

Once Alphabet has built enough production capacity for itself, the contract with SpaceX can be terminated. This could bring SpaceX some excellent short-term revenue, but it doesn't indicate that it has a long-term competitive advantage in cloud computing.