Beyond the Shock of the “0 Billion SaaS Wipeout”
“The AI That Wiped Out $270 Billion in SaaS Market Cap.” This sensational phrase was the title of a recent seminar. It symbolizes how the rise of generative AI is posing a fundamental question to the value of traditional SaaS (Software as a Service) businesses. Meanwhile, the “implementation” of AI is beginning to materialize across industries, from AI efficiency support plans for professional service firms to the inclusion of educational institution systems in AI subsidy programs.
Juxtaposing these two trends reveals the core question facing business leaders today: “What are we trying to buy with AI?” Is it AI as a mere “operational efficiency tool,” or AI as a “management resource” that transforms business decision-making and reproducibility itself? Misjudging this distinction risks turning massive investment into mere temporary efficiency gains that fail to build long-term competitiveness.
What AI Threatens is Not “SaaS” but “Mindless SaaS Dependence”
The phrase “$270 billion SaaS wipeout” evokes a simplistic narrative that AI will render all existing SaaS obsolete. But is the “form” of SaaS itself truly under threat? Rather, I believe the true threat lies in the new form of “mental stagnation” born from SaaS adoption.
In many companies, numerous SaaS applications—CRMs like Salesforce or HubSpot, communication tools like Slack or Teams, accounting software—have been adopted departmentally, effectively “outsourcing” parts of operations. This was indeed an effective way to control initial costs and quickly gain specialized functionality. However, a consequence has been that the core managerial responsibility of “designing and understanding our own business processes and data flows” has been delegated to SaaS vendors, becoming obscured.
Generative AI has the potential to directly dissect these “black-boxed business processes.” For instance, if AI agents emerge that can cross-analyze customer data scattered across multiple SaaS platforms and propose sales strategies, the previous value derived solely from the “usability” or “reporting functions” of individual SaaS tools becomes relativized. What AI truly threatens is not individual SaaS vendors, but the “management mindset that has completely outsourced business design to tools and failed to cultivate the core of its own decision-making.”
The AI Transformation of Professional Service Firms Exposes the Trap of “Routine Work”
The news about support plans launching to streamline routine tasks for professional service firms using generative AI illustrates a typical pattern in AI adoption. Contract review, articles of incorporation drafting, preparation of various application forms—these are indeed areas where AI excels, promising significant time savings.
However, the question leaders (in this case, firm principals) should ask is, “What lies beyond streamlining routine work?” If the goal stops at the dimension of “reducing labor costs,” it is merely an extension of cost-cutting “administrative IT.” The true value lies in the strategy for reinvesting the human resources freed up by AI.
For example, if AI speeds up contract review, the saved time could be used to develop a consulting service that analyzes the business risks of client companies more deeply. Or, by delegating mass case law analysis to AI, it might become possible to offer legal services in new, previously unaddressed areas. The success of AI adoption depends not on efficiency gains themselves, but on the managerial judgment of how to strategically use that “surplus.”
Do “AI Subsidies” Absolve Management of Judgment?
The news that an international student management system has been continuously registered as an eligible tool for the “2026 Digitalization & AI Adoption Subsidy” clearly shows the government’s stance of promoting AI diffusion. Subsidies certainly lower the initial hurdle for adoption and serve as a crucial catalyst for experimentation.
However, as the history of IT and DX shows, subsidies come with a significant “side effect.” There is a risk that external factors—”because we can get a subsidy” or “because competitors are doing it”—become the central motivation for adoption, while the essential internal purpose of “solving our own business challenges” fades into the background. The selection of subsidized tools often becomes vendor-led, potentially skipping the crucial process of deeply defining the “purpose” of optimal AI utilization for one’s own company.
Leaders must recognize that subsidies reduce the cost of the “means,” but do not define the “purpose.” In the case of the student management system, what is the “purpose” of using the subsidy to adopt AI? Is it merely to reduce administrative tasks, or to build the “core of a new educational service” that analyzes individual student learning data to design personalized optimal educational programs? This disparity in purpose will create a decisive gap in competitiveness three to five years from now.
What Management Should Buy is Not “AI Features” but the “Right to Redesign Decision-Making”
Summarizing the discussion so far clarifies the critical fork in the road that leaders must discern in the current wave of AI adoption. It is this: “Should we buy AI as a ‘feature’ that automates specific tasks within business processes, or invest in AI as the ‘right to redesign’ that changes the very quality and speed of the company’s decision-making?”
The former approach involves licensing general-purpose tools like ChatGPT for Enterprise or Microsoft 365 Copilot to enhance employee productivity. It certainly offers quick results. However, this is merely an improvement “along the extension of existing business.”
The latter approach is more ambitious. It aims to build specialized AI models or agents that leverage a company’s unique accumulated data (customer interactions, manufacturing logs, service records, etc.) to support its core decision-making (e.g., which customers to prioritize, how to optimize inventory, how to decide on new product concepts). The investment required here goes beyond tool licensing fees; it is an investment in internal resources for “defining the logic of the company’s own decision-making and preparing the data to train it.”
The implication of the “$270 billion SaaS wipeout” is the potential threat to the future stock price of companies that do not take the latter approach. When the source of a company’s competitive advantage becomes merely a combination of general-purpose SaaS tools available to anyone, its value is greatly diminished. Conversely, companies that possess a “reproducible blueprint” for AI utilization deeply rooted in their own business domain and data can become relatively free from price fluctuations of general tools and vendor dependence.
The Choice Facing Leaders: “AI Strategy Office” or “AI Dependence”?
So, what should leaders do concretely? The first step is to not dismiss AI as a “technical issue to be left to the IT department or information systems.” Just as cloud and SaaS adoption did before, AI is directly linked to the core of managerial judgment.
I propose the following three concrete actions:
1. Hold a meeting to define not “what we want to achieve with AI,” but “what is the highest-value work that remains for humans after delegating to AI.”
Bring together the management team and department heads to categorize the company’s work into: “routine tasks replaceable by AI,” “judgment tasks for collaboration with AI,” and “creative/relationship-building tasks exclusive to humans.” Without this discussion, appropriate investment priorities cannot be set.
2. Before applying for subsidies, ask, “Would we do this even if the subsidy were zero?”
Evaluate potential AI projects based purely on the company’s own return on investment (ROI), not on calculations predicated on receiving a subsidy. This filters for truly valuable projects rooted in solving intrinsic business challenges.
3. Start the first project from “Management IT.”
Before distributing Copilot licenses to all employees (a business/administrative IT mindset), start small with an AI application as “Management IT,” such as building an AI dashboard to support executive committee decision-making. For example, create an environment that automatically collects and analyzes management KPI data from multiple SaaS and core systems, allowing leaders to ask in natural language, “What was the main cause of last month’s sales decline?” This allows leaders themselves to experience how AI can function as a “decision-making apparatus.”
Conclusion: AI Does Not Permit “Delegation” by Management
The proliferation of SaaS, in a sense, stripped leaders of part of their responsibility for “business design,” allowing delegation in the form of “convenient tools.” However, the essence of generative AI lies in making this very “delegation” difficult. AI demands that the “underlying assumptions and logic of business processes,” previously embedded in tools, be made visible and redefined. Why is that document necessary? Why does that approval flow exist? Why do we collect that data? Without confronting these questions, one cannot simply ask AI to “make it more efficient.”
The phrase “the AI that wiped out $270 billion in SaaS market cap” signals that the market is beginning to put an end to the old model of “mindless tool dependence.” What tomorrow’s leaders should buy with AI is not a specific feature, but the “opportunity” and “right” to redesign their company’s decision-making and operational reproducibility from the ground up. Whether or not they have the resolve for this will separate the winners from the losers in the next decade.
Comments