By Javier Surasky
.png)
Developing models and building the infrastructure AI requires means committing resources whose returns depend on markets that are still taking shape. Those able to sustain such investments therefore influence which projects move forward, how quickly they do so, and under what conditions.
SoftBank
Group allows us to bring this dimension into the power map: the capacity to
mobilize capital and influence the direction in which the field develops.
SoftBank is
a Japanese group founded by Masayoshi Son in 1981. It began as a software
distributor and expanded into internet services, telecommunications, and
technology investment, including, along the way, its investment in Alibaba, the
acquisition of chip designer Arm, and the creation of the Vision Funds.
Today,
SoftBank describes itself as a strategic investment holding company, placing AI
at the center of its growth strategy (SoftBank Group Corp., 2026d).
To
understand how it operates, it is useful to distinguish between SoftBank Group,
the investment parent company, and SoftBank Corp., which operates, among other
businesses, its Japanese telecommunications activities. Although both
participate in the group’s AI strategy, they do so from different positions:
the former mobilizes investment and manages equity stakes; the latter
contributes commercial and operational capabilities.
After
examining G42, SoftBank allows us to see another way of building power through
the articulation of resources: its investments connect model developers with
semiconductor companies, sovereign wealth funds, and financial institutions,
helping to determine which technological bets receive the resources they need
to grow—and which do not.
Capital as a Capacity to Shape Direction
A company
may have researchers and a technological proposition, but it still needs
resources to sustain its development. Access to financing makes it possible to
hire staff, acquire computing capacity, and maintain operations during periods
when revenues remain insufficient.
The
allocation of capital therefore changes the possibilities available to those
competing to develop and commercialize AI.
The
relationship between SoftBank and OpenAI shows the scale this mechanism can
reach. On October 1, 2026, the Japanese group announced that it had completed
an additional USD 30 billion investment through SoftBank Vision Fund 2,
bringing its cumulative investment in OpenAI to USD 64.6 billion, equivalent to
an approximately 13% stake (SoftBank Group Corp., 2026b).
An
investment of this magnitude makes SoftBank an important partner in OpenAI’s
expansion. But its minority stake does not give it control over the company,
nor does it allow us to infer influence over specific decisions. Instead, it
reveals a relationship of reciprocal dependence in which SoftBank helps finance
OpenAI’s growth while committing resources whose future value is tied to
OpenAI’s performance.
The
relationship also has a commercial dimension. In November 2025, SoftBank Corp.,
SoftBank Group, and OpenAI announced the launch of SB-OAI Japan, a joint
venture created to offer enterprise solutions based on OpenAI technology. It
plans to commercialize Crystal intelligence in Japan and use SoftBank Corp. as
its first customer to build implementation expertise (SoftBank Corp. et al.,
2025).
This
relationship shows how investment can be complemented by facilitating access to
markets. Companies within the group contribute commercial channels and
operational experience that can make it easier for local actors to adopt
particular solutions. But this also carries a risk: if those integrations
become deeply embedded, they may raise SoftBank’s costs of switching providers
and deepen its technological dependence on its partner.
The Relationships That Sustain Its Position
The
Stargate project gave SoftBank’s financial role greater visibility. Announced
in January 2025, the project set out an ambition to invest USD 500 billion over
four years in AI infrastructure for OpenAI in the United States. SoftBank and
OpenAI were identified as the principal partners, with financial and
operational responsibility respectively; Oracle and MGX were among the initial
investors, while Arm, Microsoft, and Nvidia were among the technology partners
(OpenAI & SoftBank, 2025).
The
announced figure represents an investment intention for the project, not funds
already disbursed or contributions made exclusively by SoftBank. Still, it is
useful because it reveals a division of functions: financing, developing
models, and providing infrastructure are capabilities that must be coordinated
among actors with their own distinct interests, even when all of them share an
interest in the project’s success.
SoftBank
also owns around 86% of Arm’s equity (as of May 21, 2026), giving it a form of
corporate influence different from that derived from its minority stake in
OpenAI (Arm Holdings plc, 2026). This shows how the group combines controlling
interests, minority investments, and alliances for specific projects.
Each type
of relationship gives it different capabilities and constraints. Including
SoftBank in the power map therefore requires attention to those differences,
rather than treating every company connected to SoftBank as if it followed the
same investment logic.
Its ties
with Gulf sovereign wealth funds predate the rise of generative AI. In May
2017, SoftBank announced a first major closing of the Vision Fund with more
than USD 93 billion in committed capital, with contributors including Saudi
Arabia’s Public Investment Fund, the UAE’s Mubadala, other technology
companies, and SoftBank itself (SoftBank Group Corp., 2017). This earlier
experience points to a capability that remains relevant today: bringing
together resources from investors with different priorities and channeling them
toward technology companies.
The Conditions of Financial Power
SoftBank
also depends on others to finance its bets. In March 2026, it announced a USD
40 billion bridge financing facility intended to support its additional
investment in OpenAI and other corporate purposes. Participants included
JPMorgan Chase, Goldman Sachs, Mizuho Bank, Sumitomo Mitsui Banking
Corporation, and MUFG Bank, illustrating SoftBank’s ability to mobilize major
players in the financial sector (SoftBank Group Corp., 2026a).
In
September of the same year, the group announced approximately USD 11.1 billion
in foreign-currency-denominated senior notes, intended primarily to finance the
final tranche of its investment in OpenAI. The dollar-denominated bonds carry
annual interest rates ranging from 8.62% to 9.75%. In an October announcement,
SoftBank confirmed that this objective had been achieved (SoftBank Group Corp.,
2026c).
This
sequence allows me to identify a tension: investments rest on expectations of
future returns, while debt creates obligations with defined interest payments
and maturities. The speed at which profitability is achieved therefore becomes
important, creating an incentive for “rapid-development AI” that may come at
the expense of safer AI.
Financial
power, then, has a relational dimension. SoftBank needs to preserve access to
those who finance its own operations, maintain sustained—and ideally
rising—valuations for the companies in which it invests, and secure credit as
cheaply as possible if it is to sustain its investment capacity without major
constraints. This is a risk arising from the way SoftBank is embedded in the
ecosystem, not a prediction of crisis.
In
conclusion, the SoftBank case shows how financial decisions intervene in the
construction of technological trajectories and how those who provide capital
can expand a project’s possibilities, accelerate its growth, and help one
solution gain greater presence relative to others.
For AI
governance, this requires us to examine which priorities guide these
investments, who participates in making those decisions, and how their benefits
and risks are distributed, because the future of AI also takes shape when
someone decides what is worth financing—and for how long.
Basic Facts
- SoftBank Group Corp. is a publicly traded strategic investment holding company founded on September 3, 1981.
- It is headquartered in Tokyo, Japan, and its founder, current Chairman and Chief Executive Officer, is Masayoshi Son.
- The SoftBank Vision Funds segment had deployed USD 191.6 billion by the end of the fiscal year that closed in March 2026. This figure includes investments across different sectors and is not limited to AI.
- It maintains strong relationships with OpenAI, Arm, Oracle, Microsoft, Nvidia, MGX, PIF, Mubadala, and other financial institutions.
- Its strategic narrative includes the ambition of building a group capable of enduring for 300 years. This is a corporate aspiration, not a verifiable forecast.
References
Arm
Holdings plc. (2026). Annual report for the fiscal year ended March 31, 2026
[Form 20-F]. https://investors.arm.com/node/8281/html
OpenAI,
& SoftBank. (2025, January 21). Announcing the Stargate Project. https://openai.com/index/announcing-the-stargate-project/
SoftBank
Corp., SoftBank Group Corp., & OpenAI Group PBC. (2025, November 5). The
SoftBank Group and OpenAI launch “SB OAI Japan” joint venture. https://group.softbank/en/news/press/20251105
SoftBank
Group Corp. (n.d.). Corporate data. Retrieved October 15, 2026, from https://group.softbank/en/about/profile
SoftBank
Group Corp. (2017, May 22). SoftBank Vision Fund’s first major closing. https://group.softbank/en/news/press/20170522
SoftBank
Group Corp. (2026a, March 27). Execution of bridge facility agreement
primarily for the follow-on investments in OpenAI. https://group.softbank/en/news/press/20260327
SoftBank
Group Corp. (2026b, October 1). Execution of follow-on investment (third
tranche) in OpenAI. https://group.softbank/en/news/press/20261001
SoftBank
Group Corp. (2026c, September 24). Issuance of foreign currency-denominated
senior notes. https://group.softbank/en/news/press/20260924
SoftBank
Group Corp. (2026d). Who we are. In SoftBank Group report 2026. https://group.softbank/en/ir/financials/annual_reports/2026/who_we_are