The connection between "advanced AI, massive capital, and existing industries" is beginning.

The establishment of an AI joint venture by giant investment funds, including Anthropic (*1) and Blackstone (*2), should not be viewed simply as a "business supporting AI implementation," but rather as an example that suggests a change in the structure of corporate competition in the age of AI.

What is particularly noteworthy this time is that "advanced AI companies," "large corporations," and "existing industries" are beginning to connect at a closer distance than ever before.

Private equity funds (PE funds) like Blackstone and Hellman & Friedman (*3) are not merely investors. They are entities that implement management improvements for numerous portfolio companies. With advanced AI companies like Anthropic directly connecting to them, AI appears to be beginning to play a role closer to "management infrastructure" than just a tool for improving operational efficiency.

*1 Anthropic: A leading US AI company. Developer of the conversational AI "Claude". *2 Blackstone: One of the world's largest US investment funds. *3 Hellman & Friedman: A major US private equity fund. *4 Private equity fund: An investment company that invests in companies to improve management and increase corporate value.

AI agents are shifting from "support tools" to "task execution entities."

Traditionally, the use of AI in many companies has progressed in the form of "partially introducing AI into existing operations." However, the changes that are now beginning to occur are even more profound.

AI agents (*5) are no longer merely support tools; they are beginning to move closer to being entities that perform tasks themselves.

In particular, "management" and "production" tasks that are completed solely on a PC screen and the internet are highly compatible with AI agents. Many intellectual tasks, such as document creation, analysis, marketing operations, legal support, accounting support, programming, and customer service, are already beginning to be replaced or integrated by AI.

Furthermore, in the future, there is a possibility that we will shift from a standalone AI to a structure where "AI agents collaborate to carry out tasks."

*5 AI agent: An AI that autonomously carries out tasks based on human instructions.

The role of SaaS is changing – from "a user-operated interface" to "a protocol used by AI."

This change may also alter the very definition of SaaS (*6).

Traditional SaaS models were based on a structure where development and maintenance costs were distributed by sharing the same software among multiple companies. However, with improvements in AI-driven code generation and system building capabilities, and the development of standardized connectivity layers like MCP (*7), it will become possible to build customized systems for each company at a low cost.

In fact, Anthropic announced "Claude for Small Business" in May 2026. This is not just a chat AI, but a configuration that connects Claude to a suite of business tools that small and medium-sized businesses use on a daily basis, such as QuickBooks (*8), Google Workspace (*9), HubSpot (*10), PayPal (*11), and Microsoft 365 (*12), to provide cross-functional support for finance, sales, marketing, human resources, customer service, and more.

The key point to note is not that we are "using AI," but rather that "AI is beginning to understand business operations across the board."

Traditional SaaS was based on the assumption that humans would operate the screen. However, if AI agents begin to understand the business workflow itself and utilize necessary SaaS and external systems across platforms, the role of SaaS itself may change.

This could be seen as a sign that leading AI companies are moving beyond simply providing general-purpose AI platforms and are beginning to venture into areas closer to "operating systems for enterprises."

In that case, SaaS may move away from being a "business interface operated by humans" and towards becoming a "social common protocol used by AI agents (*13)."

In particular, legal, tax, accounting, labor, CRM (*14), and ERP (*15) systems, while seemingly complex, are actually "common social protocols" that have a set input format, approval flow, legal framework, and output format.

If AI agents begin to understand these things across the board, it's possible that entities like "SIer agents (*16)" will emerge that not only operate existing SaaS but also design and generate the necessary business systems themselves.

In other words, the "business processing layer" that SaaS has handled until now may begin to be restructured by AI agents.

*6 SaaS: Software service used via the internet. *7 MCP: A standardization concept and connection protocol for connecting AI with external systems. *8 QuickBooks: Accounting and financial management software from the United States. *9 Google Workspace: Google's suite of cloud tools for business use. *10 HubSpot: Sales, marketing, and customer relationship management tool. *11 PayPal: Online payment service. *12 Microsoft 365: Microsoft's suite of cloud services for business use. *13 Protocol: A mechanism for different systems to connect and communicate using common rules. *14 CRM: Customer relationship management system. *15 ERP: Enterprise Resource Planning (ERP) system that integrates and manages accounting, inventory, sales, etc. *16 SIer: A company that builds systems for businesses.

When "intelligence" becomes a fixed cost – the very foundation of small and medium-sized enterprises may be shaken.

This change could also impact corporate structure.

Traditionally, the existence of many small and medium-sized enterprises (SMEs) was based on the structure that "large corporations did not offer a competitive operational burden or management costs." Local focus, small-scale operations, and personalized service were all aspects that were only possible because of this human-centered management structure.

However, if AI agents begin to function as a means of replicating organizational knowledge, this structure itself could change.

If the knowledge of top sales representatives, top supervisors (*17), top instructors, and top customer service representatives can be transformed into AI and deployed across the entire organization, there is a possibility that "intelligence" itself will become a fixed cost. This is because once an intellectual operation is established, it will be possible to operate it in a way that can be mass-reproduced.

As a result, it may become easier for large companies to enter small-scale projects and decentralized markets that were previously considered to have an unfavorable ROI (*18) or high management burden.

As mentioned at the beginning, what is noteworthy is that in the United States, leading AI companies such as Anthropic and OpenAI (*19) are beginning to be closely connected with large corporations and established industries.

This appears to indicate that AI is beginning to be integrated at the business strategy level, rather than simply being used as an API (*20).

If this trend accelerates, it is possible that a gap will widen between companies that can deepen their collaboration with leading AI companies and those that cannot, in terms of the speed of business improvement, organizational evolution, and decision-making.

*17 SV: Supervisor. On-site management supervisor. *18 ROI: Return on investment. *19 OpenAI: US AI company that develops ChatGPT. *20 API: A mechanism for connecting different software programs.

The strength of Japanese companies lies in their "technology for handling the real world with high precision."

This could become an important issue for Japanese companies as well.

Of course, Japanese companies still have strengths. In particular, "technologies that handle the real world with high precision," such as high-precision sensors, optics, measurement, control, quality control, miniaturization, and power saving, are areas in which Japanese companies have accumulated expertise over many years.

In fact, as AI becomes more sophisticated, the important factor may shift from "how intelligent the AI ​​is" to "how accurately it can capture the real world."

AI cannot function on intelligence alone. To understand the real world, it absolutely needs input.

Sensors that capture real-world conditions, such as cameras, microphones, LiDAR (*21), temperature, vibration, pressure, location information, and biometric information, may become something akin to "sensory organs" for AI.

In particular, in the fields of Physical AI (*22) and robotics, the noise of the real world is extremely large. Imperfect conditions such as darkness, reflection, noise, crowds, deterioration, malfunctions, and temperature changes are always present.

Therefore, in the future, the key competitive factor may shift from "AI itself" to "how accurately it can acquire, understand, and translate the real world into action."

*21 LiDAR: A sensor technology that uses lasers to measure distance and space. *22 Physical AI: AI that acts in the real world, such as robots.

The challenge lies in the transition layer from "acquiring reality" to "understanding reality."

On the other hand, Japanese companies also face challenges.

While Japanese companies have strengths in sensors and hardware themselves, they are often not necessarily strong in UI/UX (*23) and software design, which involves "meaningful translation" of that information and converting it into a form that is easy for humans and AI to use.

In other words, while its "reality acquisition" is strong, its "reality understanding" design is weak.

What will become important in the future may not just be sensor companies, but rather integrated layers that "capture, understand, and translate the real world into action."

 *23 UI/UX: Design of ease of use and user experience.

The ultimate competitive advantage will be "the ability to implement a vision."

And beyond that comes the competition of "vision implementation ability."

Generative AI is rapidly reducing the cost of "how to create." Many of the processes involved in creating content—including text, images, videos, UI, code, and spatial design—are being accelerated by AI.

In other words, in the future, "implementation capabilities" themselves may no longer be a differentiating factor.

Ultimately, the most important question is, "What are we aiming for?"

What kind of society do you want to create? What do you find unnatural? How do you perceive humanity? What future do you want to realize?

The era is approaching where AI can rapidly materialize ideas and concepts themselves. In that sense, future competitive advantages may shift from mere technological prowess to the very vision of "what future we are trying to implement."

The changes that are beginning to occur now may be the start of a transitional process.

Author: Haruto Fujinaga