AI Transformation That Accelerates Value Creation [Finance & Implementation] — From Reducing Costs to Creating Assets: Structural Shift to Prevent ROI Loss

Insight
Feb 13, 2026
  • Management Strategy/Reformation
  • AI
GettyImages-1415118219

In recent years, many Japanese companies have proudly shared achievements as a result of introducing generative AI, such as ‘We have streamlined company-wide operations by 30%’. However, when taking a closer look at their financial results (Profit and Loss Statement, P/L), the majority of these companies have a lower SG&A expense rate and show no improvement in their operating profit margin either. What on earth happened to those tens of thousands of hours they supposedly gained?

This insight clarifies the structural factors that limit AI adoption to mere cost reductions in the P/L. It also outlines CFO-focused financial strategies and execution mechanisms to transform fixed human resources into future cash flow-generating assets.

Reference: AI Transformation That Accelerates Value Creation From Trade-Offs to Trade-Ons: A Roadmap for Genuine AI Utilization and Transformation

About the Author

  • Kinya Fujita

    Principal

Why does the ROI from AI investment disappear?

The backdrop to this issue is that a no-lay-off employment practice exists at Japanese companies. Therefore, even if these companies make business operation cuts with AI, they don’t lead directly to cost reductions (profit in the P/L statement). The newly freed-up time accumulates as “invisible idle time,” and as Parkinson’s Law suggests, it ultimately gets sucked up by new administrative tasks or meetings with unclear objectives. This is the reason why the ROI from AI investment disappears.

 Massachusetts Institute of Technology (MIT) professor Daron Acemoglu defines automation that displaces human workers with machines without delivering significant improvements in productivity, efficiency, or service quality as “so-so automation.” His latest research in 2024 warns that the impact of automation on productivity gains is extremely limited *1. Also, an external survey conducted in 2025 across five major countries shows that Japanese companies achieve lower AI adoption benefits compared with other countries *2. This is because many Japanese companies get stuck at “so-so automation”—unable to reform their actual business model.

What’s more, the AI licensing cost increases, while the labor cost remains the same. As a result, some Japanese companies actually record a worse profit margin by introducing AI.  Breaking through this problem requires a dynamic balance sheet restructuring that moves beyond the limits of short-term P/L management. Instead, use AI to force greater fluidity in fixed human resources, transforming resources into “assets” that generate future value (see Figure 1).

The essence of an AI transformation is not to simply make business operations easy. It is, ultimately, a “value creation” process of turning financially and physically transforming vanishing costs (OpEx) into assets (CapEx) that generate future cash flow.

Figure 1. Transformation From “So-So Automation” to “Dynamic Balance Sheet Restructuring”

*1: MIT professor Daron Acemoglu argues that as long as today’s AI is primarily applied to automating existing tasks, the macro-level productivity (TFP) increase could be limited to roughly 0.55–0.71% over a decade (2024). Professor Acemoglu and co-authors also identify technologies that replace workers with minimal productivity and service quality improvements as “so-so technology.” He explains the structure behind why excessive automation falls short of expectations (2019).

Three digital asset classes to invest in and leading cases

Where exactly will reallocating freed-up resources (labor costs) lead to assets that can be recorded on the balance sheet? At ABeam Consulting, we define the following three digital asset classes as investable assets that go beyond simple R&D.  This is not just armchair theory. These are assets that the leading companies featured in the previous Insight have actually built and now leverage as a source of competitive advantage.

1. Creating structural assets from implicit knowledge (proprietary AI model)

Know-how, such as the intuition and experience of expert employees, is a flow of knowledge that is lost when an experienced employee leaves. The true value of transforming this know-how into an AI model is to extract the expertise and establish it as a fixed company-owned asset.

A good example of this is from Toyota Motors. Instead of simply reducing its headcount, Toyota Motors reinvested the 10,000 hours freed up annually into “Democratizing AI.” By establishing a system in which its workers make improvements, known as Kaizen, by utilizing AI themselves, implicit knowledge within the workplace is accumulated as organizational knowledge within a proprietary model. This knowledge is then converted into development agility (see Figure 2).

2. Turning customer touchpoints into an asset (customer touchpoint asset)

This is the process of turning your engine—the core mechanism that delivers your unique CX—into assets that do not rely on external SaaS. These assets are growth infrastructure that directly raise your future top line.
Classic examples of this are from Hitachi and Schneider Electric. Hitachi (Omika Works) took its energy-saving and efficiency expertise developed at their factory and transformed it into Lumanda—a customer co-creation platform that can be sold to its customers. This is an example of a solution that goes beyond simple in-house cost-cutting.  This initiative turned a factory, which is conventionally a cost center, into a profit center that earns foreign currency. At the same time as transforming its portfolio, which redefined and integrated fragmented OT/IT businesses into digital solutions, Hitachi established a revenue base that has grown the entire Lumada business to around three trillion yen in sales (performance data from FY2024).

3. Turning data into an asset (data infrastructure)

In the construction industry, this is the equivalent of building a factory. Data is production infrastructure in the digital age. Siemens (Amberg/Chengdu Factory) has gone beyond simply improving productivity to invest in building assets in the form of a complete digital twin. By replacing physical trial and error (which wastes time, effort, and cost) in the real world with zero-cost virtual simulations, Siemens increased its production capacity by 70% while significantly reducing its marginal cost.

In order to be recognized as “assets,” they need to meet the following three criteria.

  • Reusability: It can be reused in multiple departments/processes (avoids one-off solutions created in isolation).
  • Unit economics: The unit cost decreases the more it is used (includes inference/operation cost).
  • Value alignment: It links to management KPI, such as sales, gross profit, quality, lead time (its value can be measured).
Figure 2. Case Studies of Manufacturing Sectors Realizing Trade-Ons Through AIX

Asset-generation mechanisms: Efficiency dividend and strategic investment hub

This asset conversion cannot be achieved by leaving its success to the discretion of those on the ground. Asset conversion requires strong financial governance and an implementation function to support it.

1. Financial governance: Efficiency dividend and gain sharing

Many companies think that they should do new things with the time freed up from efficiency gains. Workplaces engaged in daily operations, however, are reluctant to readily let go of resources that have been cut. What they should introduce is an efficiency dividend used by the Australian government and other similar entities in budget management.
The efficiency dividend is a mechanism whereby the headquarters preemptively recoups the expected savings upfront based on efficiency gains through investment in technology. Specifically, a set percentage (e.g., 20%) of the expected reduced work hours is forcibly recirculated back as a “dividend” to the headquarters at the start of the fiscal year by the departments that deploy AI.
One-way budget cuts, however, pose the risk of front-line burnout and superficial compliance with inward resistance. To avoid creating a “success-is-negative” culture whereby employees hide their achievements out of self-preservation, an agreement is needed where everyone shares the gains when improvements or savings are achieved.

  • Headquarter’s dividend (X%): Collect as a management resource and reinvest back into the company-wide strategy.
  • Retain at ground level (X%): Allow as a future investment bracket (R&D and new business budget) that departments can use freely.

For example, let’s assume a company with a 500-strong back-office function expects to cut work hours by 20,000 hours annually using generative AI. If an efficiency dividend of 20% is applied at the beginning of the fiscal year, the headquarters can recoup the equivalent of 4,000 hours. The strategic investment hub then invests 50% of the recouped work hours into cross-department assets (shared knowledge, data infrastructure, reusable components), while the other 50% is recirculated as a future investment bracket for the department. The key point here is that the recouped time is reliably fed into the investment pipeline for asset generation and not used up by additional operations.

2. Implementation functions: Strategic investment hub and platform engineering

In the previous Insight, we proposed establishing an AI Center of Excellence (AI CoE) as the central command hub. Rather than simply operating as a technical support team (cost center), it must function as a strategic investment hub (profit center) that turns recouped resources into assets.
The true role of the strategic investment hub is to rigorously review proposed initiatives to determine whether they are one-off efficiency tweaks or reusable, long-term assets, and to invest only in initiatives that qualify as assets by meeting the criteria for reusability, unit economics, and value alignment.
What’s essential here is applying the platform engineering adopted by tech companies to management. This originates from the “Paved Road” concept put forward by Netflix. It’s important to note that this is by no means a concept exclusive to tech companies. Netflix was originally a logistics and rental company that sent out rental DVDs. In the process of transforming its business from a physical operation to a digital operation, Netflix created a standard path for its developers to focus on value creation, dramatically increasing its productivity.
Applying this thinking to corporate management, the key to asset creation is building a “paved road” that enables safe and quick development by all employees, focused on the strategic investment hub (see Figure 1, Figure 4).

Figure 3. Example of Applying Platform Engineering at an Industrial B2B Company
Figure 4. Asset-Creation Process Through Efficiency Dividend and Strategic Investment Hub

Redefining operating leverage and human resources that enable trade-ons

There is a trap here, however, that CFOs must not fall into.  Simply reclassifying labor costs as software spend risks bloating the balance sheet and actually lowering the price-book-value ratio (PBR).
What investors value is not simply the sheer amount of assets. It’s whether these assets actually offer an operating leverage (see Figure 5). Operating leverage is a metric that measures the sensitivity of profit to changes in revenue. It is expressed as (revenue − variable costs) ÷ operating profit. This formula highlights the steadfast rule that companies should minimize variable costs (maximize the contribution margin). What’s vital here, however, is how labor costs are seen at Japanese companies.

  • OpEx (personnel) model: Labor costs are classified as fixed costs for accounting reasons. However, in labor-intensive models, increasing revenue requires a roughly proportional increase in headcount. In other words, structurally, because labor costs behave like a variable cost, it makes profit margin improvements difficult, even at scale.
  • CapEx (AI asset) model: Meanwhile, the marginal cost (variable cost) of AI and data platforms that have been made into assets is almost zero. As the cost (asset) does not increase even when sales do, the contribution margin increases in proportion to revenue, resulting in a dramatic rise in operating leverage.
Figure 5. Achieving Operating Leverage (Trade-Ons)

Operating leverage represents the financial essence of a “trade-on”. That is, achieving both growth and efficiency. The more the business grows, the lower the share of AI asset depreciation becomes, and profit margins improve exponentially. This shift to a non-linear revenue structure is the goal of dynamic balance sheet restructuring.
Increasing operating leverage means accepting the risk of profit deterioration (volatility) during periods of revenue decline. The textbook financial response to uncertainty is to reduce the break-even point by making costs more variable, such as through outsourcing.

However, in Japan, where the labor supply is shrinking and labor costs continue to rise, continuing to rely on personnel is precisely what could be a medium- to long-term management risk. Labor costs are an “inflationary asset” that will continue to rise. Meanwhile, AI and computing costs are a “deflationary asset” that will continue to reduce in cost relative to performance. The shift is to change inflationary variable costs (personnel) to deflationary fixed costs (AI). Generally, changing to fixed costs tends to be seen as a risk, but it is actually a rational strategy during periods of inflation. Shifting the cost structure to company-owned assets that are less affected by external factors like rising wages is what will minimize volatility and act as true governance to ensure management stability.

In an age of population decline, labor will become an extremely rare resource. Restricting your valuable staff to processing tasks that could be done by AI is a managerial loss.  The role of personnel in an AI transformation will be redefined in the following two ways.

  • AI supervisor: Staff who audit and correct AI output and are responsible for new routine work to ensure quality
  • High-touch: Customer-facing and field operations which add value because human presence and interaction cannot be replicated by AI

While AI assets with a near-zero marginal cost ensure efficiency at scale, people create value in uniquely human areas involving qualitative and interpersonal work. This redesigning of the division of roles is an inevitable strategy to survive this age of labor shortages.

As we have explored so far, an AI transformation is not just a case of deploying tools. An AI transformation is a fundamental shift of the company’s management operating system (core foundation) from a labor-intensive profit/loss mindset to an asset-based balance sheet mindset. Its success no longer rests solely with the CIO or CTO. It lies with the CFO and CHRO, who govern the flows of people and capital. The topic of discussion in management meetings should not be ‘how can we cut work hours’. Neither is it approving a perfect roadmap drawn up by an outside consultant. What should truly be up for discussion are three questions: ‘What percentage of the company’s resources can the CFO reclassify from costs through efficiency dividends and gain sharing?’ ‘Which assets will they be converted to?’ and ‘How will they contribute to improving the PBR (operating leverage)?’
An AI transformation is a structural shift process to restructure the rigid balance sheets at Japanese companies. The key to its success and whether engineers’ innovations are turned into financial assets ultimately depends on the CFO’s leadership and decisions.

ABeam Consulting will continue to seek to help clients accelerate value creation by leveraging AI and achieve speedy and reliable business transformation through the provision of consulting services aligned to each individual company’s goals and stages of development.

References

Insights

Contact

Click here for inquiries and consultations