Key Points:
- Global artificial intelligence infrastructure will require an annual investment of $5 trillion by 2040.
- AI revenue is projected to generate over 20% of global gross domestic product, representing $25 trillion annually.
- The computing network will demand 3 terawatts of electricity, equivalent to 1.8 times current global consumption.
- By 2040, the global ecosystem will host over 100 trillion autonomous AI agents making independent decisions.
The unprecedented race to construct artificial intelligence infrastructure will soon demand capital on a massive scale, transforming the entire global economy. During a major mid-summer corporate address in Tokyo, SoftBank Group chief Masayoshi Son laid out a massive vision for the future of technology, projecting that global AI buildout costs will climb to $5 trillion annually by 2040. This immense investment, equivalent to roughly 800 trillion yen each year, will fund the construction of massive data centers, energy networks, and advanced robotics pipelines to transition the world away from human-centric operations.
This bold capital projection arrives amid a growing debate on Wall Street regarding whether current tech valuations have entered bubble territory. However, Son dismissed these technological bubble concerns as completely absurd, asserting that critics who question the sustainability of the boom do not understand the true capabilities of superintelligence. While skeptics worry about short-term returns on capital, the underlying economic value created by automated agentic systems will quickly make today’s multi-billion-dollar investments look like minor startup expenses.
The financial viability of this multi-trillion-dollar annual expenditure relies on a fundamental transformation of global productivity. By 2040, artificial intelligence applications and services will generate more than 20% of global gross domestic product (GDP), representing roughly $25 trillion in annual economic output. Within this highly automated economic landscape, spending $5 trillion annually on infrastructure stands out as a highly defensible and necessary capital expenditure. The massive revenues generated by these digital systems will easily absorb the immense manufacturing and construction costs.
This economic transformation will occur as the technology progresses past standard large language models and into the era of artificial superintelligence (ASI). By 2040, the global ecosystem will host more than 100 trillion autonomous AI agents. These digital entities will possess the capability to make independent decisions, execute complex corporate tasks, and communicate seamlessly with other automated agents. This massive network of digital intelligence will handle the vast majority of administrative, engineering, and operational workloads worldwide, systematically displacing human labor from traditional roles.
Operating such an immense computational network will require a massive overhaul of global energy infrastructure. The total electricity demand for AI-focused data centers will reach 3 terawatts by 2040, a figure representing roughly 1.8 times the current global electricity consumption. Meeting this unprecedented power demand will require a multi-stage energy strategy. While natural gas power plants and upgraded electrical grids will handle the early stages of this capacity expansion, the long-term survival of the superintelligence network will ultimately rely on the commercialization of nuclear fusion technology.
SoftBank is already executing a multi-billion-dollar strategy to position itself as the premier full-stack platform provider for this digital future. The company’s total investment in ChatGPT-maker OpenAI is on track to surpass $60 billion before the end of the year, backed by a massive $40 billion bridge loan. These aggressive capital moves follow a highly successful financial year ending March 31, where SoftBank recorded a record net profit of 5 trillion yen (approximately $32 billion), representing a massive five-fold increase from the prior year driven directly by its lucrative AI holdings.
To achieve this dominant market position, the investment group is integrating physical infrastructure, semiconductors, software models, and robotics into a singular ecosystem. The company controls a massive share of the semiconductor design market through its ownership of Arm, which recently developed its first high-performance central processing unit designed specifically for artificial general intelligence. By linking this chip-design power with its massive data center projects, the company can deploy custom hardware optimized specifically to run autonomous agent software.
The physical expansion of this computing network is already underway on a global scale. In March, the company broke ground on the Ohio Project, a massive 10-gigawatt technology campus in the United States designed to house advanced AI servers. This was quickly followed in June by the launch of the France Project, a €5 billion initiative to build 5 gigawatts of data center capacity in Europe. These megawatt-scale developments represent the initial building blocks of the massive computational grid required to support the transition to superintelligence.
The final piece of this full-stack strategy involves deploying physical bodies to execute tasks in the real world. The technology group is heavily funding advanced humanoid robotics, aiming to complete its acquisition of a major European robotics business by late 2026. By combining on-device artificial intelligence with highly advanced mechanical limbs, these humanoid robots will perform physical labor in factories, warehouses, and logistics centers. This integration of software minds and robotic bodies represents the commercialization of physical superintelligence.
Ultimately, the bold projection of a $5 trillion annual AI investment illustrates the sheer scale of the digital transition ahead. While traditional analysts struggle to reconcile these massive expenditures with short-term quarterly earnings, the architects of this technology are playing a multi-decade game to build the physical foundation of superintelligence. As the global demand for electricity, silicon, and data centers continues to soar, the companies capable of funding and securing this infrastructure will hold the keys to the future of the global economy.
