Investing.com -- SpaceX (NASDAQ: SPCX) could face costs of $50 billion to $130 billion, including spectrum, to build a standalone terrestrial network for its direct-to-device mobile business, according to Bernstein analysts, marking one of the more challenging pieces of the company’s broader satellite strategy.

Analysts led by Douglas Harned said they remain optimistic about SpaceX’s overall opportunity across launch, orbital data centers, and Starlink broadband for consumer, enterprise, and government customers, calling this combination the basis for its "positive outlook for SpaceX stock." But the analysts call the direct-to-device mobile business "the one business that we have viewed as the most difficult."

SpaceX plans to begin Starship launches of its Mobile V2 satellite constellation in mid-2027. Bernstein’s cost estimate, which ranges from roughly $15 billion to $80 billion excluding spectrum, equates to an estimated 30,000 to 120,000 macro sites, depending on network design choices.

The analysts said the wide range hinges on two key decisions — how aggressively SpaceX wants to compete with terrestrial carriers on network quality, and whether the company acquires 10 MHz of low-band spectrum that Grain plans to auction.

Elon Musk responded "not true" on X to a Bloomberg report suggesting SpaceX was eyeing the Grain spectrum, though Bernstein believes "a Grain spectrum acquisition is not off the table." AST Spacemobile (NASDAQ: ASTS) has also expressed interest in the spectrum, receiving a 30-day special temporary FCC approval in August to test supplementary coverage on it.

Bernstein modeled six scenarios based on those two variables, ranging from a bare-bones "Metro" network covering about 70% of population, similar to DISH’s build, to a fully loaded "Premium" network approaching competitiveness with existing telecom carriers. A middle "National" scenario, more comparable to Sprint’s historical network, was described as potentially credible "at the right price point."

The broker’s base case is a National build incorporating Grain’s spectrum, estimated at roughly $70 billion and 57,000 macro sites over an approximately eight-year build period. Analysts said acquiring even a small amount of low-band spectrum, such as Grain’s holdings, could reduce the number of required sites by about 30%, since Grain represents "the only obviously available low band spectrum."

Despite the detailed cost analysis, the team said it still views a partnership as the most likely outcome for SpaceX’s mobile business. However, they noted that "the company continues to indicate a terrestrial buildout is possible," prompting the detailed assessment given ongoing investor discussion of high potential costs across the telecom and tower sectors.