SoftBank's Son: AI Bubble Talk is "Blasphemy" — And He Needs You to Believe That

TL;DR

• Masayoshi Son told SoftBank shareholders at the June 24 AGM in Tokyo that any talk of an AI bubble is "an insult to AI" (Reuters, "blasphemy against AI").
• He set a net asset value target of ¥1,000 trillion ($6.189 trillion) within the next decade — roughly 40x SoftBank's current NAV.
• The target depends on "superintelligence" arriving and being monetised through SoftBank's roughly 90% stake in Arm Holdings and its $30B+ position in OpenAI.
• SoftBank shares are up ~70% year-to-date, but S&P Global revised its credit outlook to negative in March, citing deteriorating asset liquidity from the OpenAI bet.
• The two statements — "no bubble" and "¥1,000 trillion NAV" — are not independent. The NAV target requires the bubble thesis to be wrong.

What Happened

At SoftBank Group's annual general meeting in Tokyo on Wednesday, founder and CEO Masayoshi Son delivered what has become signature theatre: a shareholder address that blends visionary ambition with dismissive rhetoric aimed at skeptics.

The headline numbers:

• NAV target: ¥1,000 trillion ($6.189 trillion at current exchange rates), to be achieved "in the next decade."
• The quote: Any talk of an AI investment bubble is "blasphemy against AI" — or in Reuters' translation of Son's remarks, "an insult to AI." AI is "still in its early stages."
• Current context: SoftBank's current market capitalisation is roughly ¥16 trillion (~$100B). Its net asset value — the market value of its holdings minus liabilities — sits somewhere south of ¥25 trillion, depending on how you mark Arm's most recent private valuation. The target represents a roughly 40x increase from today.
• Ancillary announcements: Son confirmed SoftBank is building AI data centres in the U. S. and is seeking to invest in Tokyo Electric Power Co (9501. T) as it looks to bring in external capital.

The meeting also coincided with the finalisation of Hyundai Motor Group's acquisition of SoftBank's remaining 9.65% stake in Boston Dynamics for $325 million — closing out Son's robotics bet at a modest return while freeing capital for the AI push.

(Sources: Reuters, WSJ, Dow Jones Newswires — June 24, 2026.)

What It Actually Means

There are two stories here. One is about a man whose track record on bold predictions is genuinely remarkable. The other is about capital-markets physics.

The track record. Son called the internet revolution early and rode it to become Japan's richest man. He predicted SoftBank would own 90% of the global chip market (it didn't — but Arm is the closest thing to an architectural standard outside x86). He bought Sprint and turned it around. His Vision Fund pioneered the $100M+ cheque at startup stage. Son's willingness to make numerically specific, decade-ahead predictions that sound absurd at the time — and occasionally land — is part of his market power. Investors who dismiss him categorically have missed returns.

The physics problem. SoftBank's current NAV is roughly ¥20–25 trillion. To reach ¥1,000 trillion requires 40x growth. Even assuming aggressive compounding — say 25% annual NAV appreciation, which would be extraordinary for a holding company — the timeline stretches past 15 years, not 10. A 10-year target implies roughly 45% compound annual NAV growth.

The only asset in SoftBank's portfolio with the theoretical potential to deliver that is Arm. Arm currently accounts for roughly 30–40% of SoftBank's NAV. For the NAV target to be achievable without a miracle:

1. Arm would need to grow from its current ~$100B valuation to something approaching $3–4 trillion — roughly where Nvidia sits today.
2. OpenAI (another ~30% of NAV post the additional $30B investment) would need to IPO at a valuation that justifies the mark.
3. The rest of the Vision Fund portfolio — including dozens of private companies, many still pre-revenue — would need to produce multiple home runs.

This is not impossible. It is, however, priced on a probability curve that most institutional investors would describe as a small number of standard deviations from the mean.

Hype Deconstruction — What This Isn't

This is not a conventional corporate strategy announcement. The ¥1,000 trillion figure is aspirational in the way that Elon Musk's "$25 trillion Tesla valuation" prediction was aspirational — it functions as a north star for employee and investor psychology, not a financial plan that any CFO would support in a board paper.

The "blasphemy" framing is the giveaway. When a CEO responds to a question about overvaluation by invoking religious language, they are signalling that the thesis is not falsifiable from inside the building. That is a feature of successful technology founders — Bezos was similarly dismissive of bubble talk in 1999, and he was right in the long run — but it is also the same language every bubble defender has used at the peak.

The timing matters. Son made these remarks the same week the Los Angeles Times published a piece headlined "How the coming AI-related IPOs could puncture the AI bubble," arguing that insider IPO behaviour suggests "we are approaching peak AI." The contrast is the point.

Stakeholder Landscape

Cross-Layer Implications

Credit markets: The most immediate non-obvious effect. SoftBank's borrowings — including the substantial loans against its Arm stake — are priced on the assumption that NAV is manageable relative to debt service. If the gap between Son's aspirational target and actual NAV becomes a topic of analyst scrutiny, SoftBank's credit default swap spread widens. This feeds back into the cost of capital for the Vision Fund's next round of investments.

Geopolitics: SoftBank is building AI data centres in the U. S. while seeking to invest in Tokyo Electric Power. This is an explicitly geopolitical play — Japan positioning itself as a non-China, non-U. S. node in the AI infrastructure supply chain. Son's target only works if Japan remains in the top tier of AI-capable nations. If export controls on advanced chips tighten further, SoftBank's access to Nvidia hardware becomes a constraint on the data centre buildout, and the NAV target gets harder before it gets easier.

Talent markets: SoftBank's portfolio companies — Arm, OpenAI, the various Vision Fund startups — are in direct competition for AI talent. The NAV target signals to those employees that their equity is backstopped by a parent company with a 10-year horizon. That matters for retention at companies where star researchers are being poached by hyperscalers offering cash comp packages north of $5M/year.

Uncertainty Ledger

• What exchange rate is Son assuming? ¥1,000 trillion = $6.189 trillion at current ¥161.5/USD. If the yen strengthens to ¥130/USD, the dollar target drops to $7.7 trillion — a very different conversation.
• OpenAI IPO timing. If OpenAI's IPO is delayed past 2027, the NAV target becomes a 7–8 year problem, not 10. That changes the required CAGR.
• Arm's growth trajectory. Arm's licensing revenue from AI inference at the edge is real but unproven at scale. If the edge-AI thesis underdelivers, the anchor asset of Son's portfolio softens.
• S&P's negative outlook. A downgrade to junk status would trigger forced selling by investment-grade-only mandates, compressing SoftBank's capital-raising flexibility at precisely the moment the NAV target demands aggressive deployment.

What This Means for You

For retail investors who hold SoftBank or Arm: You are now invested in a fund that has declared a 40x return target over 10 years. That is a venture fund's ambition with a public-company liquidity profile. Rebalance accordingly — or accept that you own a story stock with a CEO who uses religious metaphors to deflect valuation questions.

For institutional allocators with SoftBank exposure: Your risk models for the position are now wrong. The "blasphemy" remark is not a strategy update; it is an organisational signal that Son will not tolerate internal skepticism about pricing. That increases the probability of overpayment in future Vision Fund rounds. Adjust for adverse selection.

For anyone watching the AI capex cycle: Son is the most consequential single voice arguing that AI infrastructure spending is rational at current levels. If he is wrong, the correction is not just a SoftBank problem — it is a systemic one, because SoftBank's spending has been a material contributor to AI GPU demand. If he is right, ¥1,000 trillion NAV means SoftBank will be the most valuable company in history.

For general observers: The most useful thing you can do is ignore the ¥1,000 trillion number and watch the credit markets. When SoftBank's bond yields start moving independently of Son's rhetoric, that is the signal. Everything before that is narrative.

Bottom Line

Masayoshi Son's ¥1,000 trillion NAV target and his "blasphemy" dismissal of AI bubble talk are the same statement delivered in two registers. One is a number that functions as a story. The other is a story that refuses to be cross-examined by numbers. Son has been right before when the world called him wrong — but the gap between SoftBank's current NAV and his target is so large that even his best historical performance does not bridge it without assuming that "superintelligence" arrives on schedule and is monetised at a multiple the market has never seen. That is not an investment thesis. It is a wager that history will repeat in compressed form. It might. But "blasphemy" is not an argument.

Sources:

1. Reuters — "Talk of a bubble is 'blasphemy against AI' says SoftBank's Son" (June 24, 2026) [Tier 1]
2. Wall Street Journal — "SoftBank's Son Sets Bold Net Asset Value Target on Promise of 'Super AI'" (June 24, 2026) [Tier 1]
3. Dow Jones Newswires — "SoftBank's Son Shoots for $6.189 Trillion NAV on Promise of 'Super AI'" (June 24, 2026) [Tier 1]
4. CNBC — "SoftBank's OpenAI bet and rising debt are raising liquidity crunch concerns" (June 4, 2026) [Tier 1]
5. CNBC — "Fears over SoftBank's balance sheet are overblown: Comgest" (June 2, 2026) [Tier 2]
6. S&P Global Ratings — SoftBank credit outlook revision to negative (March 2026) [Tier 1]