Softbank CEO dismisses Elon Musk’s extraterrestrial data center idea in favor of maximizing Earth-side construction now: ‘He who strikes first wins’

Masayoshi Son reiterated that SoftBank is prioritizing Earth-based AI data centers over Elon Musk’s orbital concept, arguing space deployment would take years and cost too much to be practical. SoftBank remains a major backer of OpenAI’s Stargate project, having committed $19 billion toward a $500 billion AI infrastructure buildout. The article also highlights SoftBank’s long-term goal of reaching 1 quadrillion yen in net asset value and its continued push into artificial superintelligence and robotics.

Analysis

The market is still pricing AI infrastructure as a pure compute race, but the more durable edge is likely to be whoever can lock in land, power interconnects, permits, and financing fastest on Earth. That favors incumbent cloud and infrastructure owners more than speculative orbital concepts, because the bottleneck over the next 12-24 months is deployment velocity, not theoretical efficiency. The implication is that the winners are the boring balance-sheet assets that can turn capex into capacity now, while the losers are the story stocks whose valuations implicitly capitalize a much longer-dated optionality.

For TSLA, the negative read is not the space narrative itself; it is the renewed reminder that investor enthusiasm can outrun near-term monetization. If the market starts haircutting long-duration AI fantasies, multiples for adjacent “future platform” names can compress even without any change in operating fundamentals. That creates a setup where any disappointment in autonomous/AI execution could hit harder, because the stock has been partly functioning as a call option on multiple future technology pathways.

The contrarian point is that orbital data centers are not a near-term P&L issue but a real option on energy-constrained compute in the 2030s. If launch costs keep falling and power density keeps rising, today’s skepticism could become tomorrow’s free option embedded in private-market valuations. So the right response is not to dismiss the space thesis entirely, but to price it as zero today and avoid paying for it implicitly in public comps that already assume flawless execution.

For BABA, this is mostly incidental unless the market re-rates all AI adjacency names with a higher hurdle rate for capex intensity. On balance, the article modestly supports a preference for infrastructure enablers and away from premium-duration narratives, with the biggest risk being a sudden policy or technological breakthrough that collapses the timeline and reignites multiple expansion in the highest-beta AI names.

Key Decisions for Investors

• Short-dated TSLA downside hedge: buy 1-3 month put spreads against strength in the stock; the setup favors a multiple reset more than an earnings-driven selloff, so theta should be managed tightly.
• Relative value: long earth-bound AI infrastructure beneficiaries vs. long-duration AI story names over 3-6 months; use a basket of power/grid/data-center enablers against TSLA as the cleaner sentiment-sensitive short.
• Stay constructive on BABA only on fundamentals, not AI narrative optionality; avoid paying up for any perceived AI frontier premium in the name until there is evidence of monetization.
• If seeking event-driven expression, use a TSLA call spread financed by puts only if you believe the market is underestimating a near-term AI catalyst; otherwise the skew still looks unfavorable to outright long premium.
• Maintain a watchlist for private-market infrastructure names tied to permitting, transformers, and grid equipment; they are better positioned to capture the next 12-24 months of AI capex than speculative orbital concepts.