Son Masayoshi’s AI gamble is now being called to account by the market. SoftBank Group will release its quarterly earnings tomorrow (November 10). But even before the report, its stock has suffered consecutive blows: it plunged about 11% last Friday and fell another 4.2% today, closing at 3,400 yen—a decline of nearly 23% year to date. Reuters notes that investors are no longer primarily interested in how large Son plans to build his AI empire; they now want to know exactly how he intends to build it. SoftBank has invested approximately $64.6 billion and holds a 13% stake in OpenAI, while making broad bets across chips, data centers, and robotics. Yet as of the end of June, the group’s net interest-bearing debt had surged to 10.8 trillion yen, up from 6.6 trillion yen a year earlier. As core holdings like Arm have declined in value, SoftBank’s ability to secure further financing through asset collateral has come under pressure, with credit default swap costs reaching a four-year high. Compounding the problem, just as financial pressures mount, geopolitical headwinds have emerged. China’s regulators blocked SoftBank’s $5.4 billion acquisition of ABB’s robotics business in China, halting the deal. Both parties are exploring options to spin off the Chinese operations to salvage the transaction—but China happens to be ABB Robotics’ single largest market. Son has long presented a comprehensive narrative spanning AI models, chips, computing infrastructure, and physical robots. Now he faces three walls simultaneously: diminishing market patience for fundraising, real-world constraints on AI infrastructure, and geopolitical scrutiny over technology assets. That’s why tomorrow’s earnings report may matter more than the actual profit figures. For years, SoftBank’s greatest strength was betting early on the next era and letting capital markets price in the future. But when future valuations can no longer be easily converted into today’s fundraising capacity, the investment story must now be tested by execution. The market hasn’t suddenly stopped believing in AI—it’s now asking: If AI’s future is truly worth so much, why does it still require constant borrowing just to build it?