Capital & Industrial Strategy

SoftBank Group has surpassed Toyota Motor to become Japan's most valuable company by market capitalisation as of Monday, a milestone that would have been structurally implausible even two years ago. Toyota held the top position for over 20 years, anchored by its dominance in global automotive manufacturing. The displacement reflects a broader repricing of AI-exposed holding companies — SoftBank's portfolio includes substantial stakes in Arm Holdings, whose architecture underpins much of the AI chip ecosystem, as well as a range of AI-adjacent venture positions via its Vision Fund vehicle. Bloomberg and Financial Times both confirm the market cap crossing.

The strategic implication for capital allocators is significant: Japan's equity market, long dominated by industrial and automotive conglomerates, is now pricing AI infrastructure exposure above physical manufacturing at scale. SoftBank's rise is also a proxy bet on Arm's continued royalty expansion as AI silicon proliferates across data centres and edge devices. Masayoshi Son's rehabilitation after the Vision Fund losses of 2021-2022 is now complete in market terms, though the underlying portfolio quality of Vision Fund II remains an open question.

Nvidia used a series of coordinated announcements to extend its silicon platform beyond AI data centres in three directions simultaneously. First, the Vera CPU — confirmed by CEO Jensen Huang to have Anthropic, OpenAI and SpaceX as early adopters — positions Nvidia as the compute substrate for frontier AI inference workloads, not just training. Bloomberg and Reuters confirm the customer list. Second, Nvidia has entered the Windows laptop market with an Arm-based PC chip, partnering with Dell, HP, Microsoft, ASUS and Lenovo for the first hardware generation. Bloomberg, Financial Times, WSJ and CNBC all confirm the launch. Third, Nvidia has selected Chinese robotics startup Unitree as the hardware platform for its first publicly available humanoid robotics research system, per CNBC.

The PC move is strategically distinct from Nvidia's core business: margins in consumer PC silicon are structurally lower than data centre GPU margins, and Intel has decades of OEM relationships and ecosystem lock-in. The strategic logic is not near-term revenue maximisation but platform control — by owning the silicon in AI-native PCs, Nvidia secures the edge inference layer as AI agents migrate from cloud to device. The Arm architecture choice aligns with Apple Silicon's success and reduces licensing dependency on x86. The Unitree partnership is notable given US-China technology tensions: it allows Nvidia to seed its robotics software stack globally while keeping the relationship at the research platform level rather than a direct equity investment.

A new Bain global survey of large companies finds that cost savings from AI automation are broadly missing projections, with the consultancy explicitly stating the shortfall 'should be making executives uncomfortable.' Bloomberg reports the findings. This follows a pattern visible in earlier surveys from McKinsey and Gartner suggesting enterprise AI is still predominantly in pilot or limited-deployment phases rather than at the scale required to generate material cost reduction. The gap between committed capex — particularly on cloud AI services and software licensing — and realised productivity gains is the central tension in the enterprise AI adoption cycle.

The investment implication is dual-directional. In the near term, it creates pressure on enterprise AI software vendors whose sales cycles have been built on ROI projections that are now under scrutiny. It also intensifies focus on agentic AI as the mechanism that will finally deliver automation at scale, which is partly why Nvidia's PC announcement explicitly frames the new hardware as designed for 'AI agents' rather than general AI. For capital allocators, the Bain data is a counterweight to the equity market's AI exuberance: the gap between infrastructure spending (where returns are flowing to chip and cloud providers) and enterprise value creation (where returns have not yet materialised for buyers) remains wide.

The US has moved to halt Nvidia AI chip shipments to Chinese firms operating outside China, according to Reuters. The mechanism targets Chinese entities using subsidiaries or fronts in third countries to procure restricted chips — a loophole that has been widely documented. This tightening is consistent with the strategic logic articulated in a WSJ opinion piece arguing the US should 'starve adversaries of AI compute,' WSJ, though that piece represents commentary rather than policy confirmation. The confirmed regulatory step from Reuters narrows Nvidia's effective addressable market for its highest-margin data centre products.

Simultaneously, the memory chip supply stack is being aggressively repriced by investors. A top-performing technology fund has announced plans to acquire SK Hynix shares, betting on continued HBM memory supply tightness, per Bloomberg — this is an announced intention, not a closed position. SK Hynix has already rallied approximately 1,000% over the past year. South Korea's export growth hit a four-decade high driven by AI memory chip demand, per Reuters. TSMC's local Taiwan-listed shares are narrowing their discount to the ADR, per Bloomberg, as local investors price in continued AI demand. Samsung and LG shares also rallied ahead of Jensen Huang's meetings with Korean executives, per Reuters.

Two China AI funding signals emerged Monday. Vast, a 3D-modelling startup founded by a 29-year-old, raised nearly $200 million to reach a $1 billion valuation, becoming China's latest AI unicorn per Bloomberg. Separately, MiniMax — one of China's more prominent foundation model startups — is exploring a listing on Shanghai's STAR Market, per Reuters. The STAR Market listing is an explored intention, not a confirmed filing.

The MiniMax STAR Market signal is strategically significant beyond the single company: it indicates that China's domestic public markets are being positioned as the primary liquidity pathway for AI startups as US IPO access narrows under export control and geopolitical pressure. This creates a bifurcated global AI capital market — US-listed AI companies trading at premium multiples, China's AI ecosystem increasingly funded and exited through domestic channels. LG Electronics' 300%-plus share rally on physical AI and robotics positioning, per Bloomberg, is a separate but related signal that Asian industrial companies repositioning toward physical AI are being aggressively repriced by investors.