The Era of Simultaneous AI Investment Growth and Uncertainty
Major US IT companies are reporting consecutive profit increases. The pattern of cloud demand and AI-related services boosting revenue is now standard. However, as reported by NHK News, concerns over massive investments are also growing. The uncertainty around return on investment (ROI) is a common challenge for executives.
Amidst this, NTT’s “AIOWN” initiative offers intriguing insights. The company plans to triple its domestic data center (DC) capacity by fiscal 2033 to meet growing AI inference demand. This concept, branded as AI-native infrastructure, is designed not just as a capital investment but on the premise that AI will be integrated into the core of management.
Another key point is the discussion reported by IT Leaders about “elevating AI use from task support to decision-making.” Recognizing that many companies stop at using AI as a “convenient support tool,” it outlines a path to the next phase.
This article, based on these latest news, explains specific criteria and practical methods for executives to evolve AI from “task support” to a “foundation for decision-making.”
The New Thinking on Infrastructure Investment Shown by “AIOWN”
NTT’s AIOWN is an infrastructure specialized for AI inference processing. Unlike traditional cloud services, it uses an architecture optimized for running AI models. By tripling domestic DC capacity, it aims to provide a low-latency, secure AI inference environment.
There are three key points for executives to understand.
First, the premise is that demand for AI inference will accelerate exponentially. With the spread of generative AI, diverse inference processes like image recognition, voice processing, and data analysis, in addition to text generation, will permeate daily operations. Your company’s AI utilization plans must be designed with this expansion in mind.
Second, consider the cost of infrastructure. As specialized infrastructure like AIOWN becomes common, AI processing can be executed more efficiently than on traditional general-purpose clouds. Consequently, the running costs of AI utilization may decrease. The era is approaching where you don’t need a massive in-house AI platform; you can use what you need, when you need it.
Third, the importance of data governance. Expanding domestic DC capacity addresses the need for domestic data storage. For companies handling highly sensitive data, especially in finance, healthcare, and government, this makes balancing AI utilization and data protection more realistic.
Three Steps to Evolve AI into the Decision-Making Domain
The theme of “moving from task support to decision-making,” as highlighted in the IT Leaders article, is an urgent issue for many companies. From my own experience supporting AI implementation in over 38 clients, I am confident there are clear steps to this evolution.
Step 1: Data Integration and Quality Improvement
High-quality data is essential for AI to support decision-making. A common challenge for many companies is data siloed by department. Start by centralizing the data needed for management decisions and performing data cleansing (quality improvement). Specifically, the first step is to make sales data, customer data, inventory data, and HR data interoperable.
Step 2: Building and Validating Predictive Models
Once data is in order, the next step is to build predictive models. Start with topics directly linked to management decisions, such as demand forecasting, churn prediction, or employee turnover prediction. The key here is not to aim for perfection. A realistic approach is to first test with a simple model, validate the results, and then improve accuracy.
Step 3: Integrating AI Judgments into Management Processes
The final stage is incorporating AI analysis results and predictions into actual management decision-making processes. For example, include AI prediction data as an agenda item in monthly management meetings, or reference AI simulation results during budget planning. The crucial point at this stage is not to treat AI as a “black box.” Choose AI that can explain its reasoning, or use mechanisms to enhance explainability.
Insights from the “Survey on How AI is Changing Workplaces and Management Tasks”
A survey published by Japan’s HR department is also noteworthy. This survey quantitatively analyzes the impact of AI utilization on workplaces and management. While specific figures aren’t detailed in the article, the survey’s framework itself contains important implications.
As AI adoption progresses, the role of management shifts from “giving instructions and orders” to “designing and evaluating AI utilization.” In other words, the skill set required of managers fundamentally changes.
In my experience, in companies where AI implementation is successful, management is actively enhancing their AI literacy. Conversely, in companies where AI adoption is slow, managers often fear that “AI will take away jobs.” This difference in mindset determines the success or failure of AI utilization.
Three Actions Executives Should Start Today
Theory alone won’t move the organization. Here are three concrete actions executives can take starting today.
Action 1: Inventory Your Decisions
Classify your company’s management decisions into “routine decisions” and “non-routine decisions.” Routine decisions (inventory ordering, price adjustments, shift scheduling, etc.) are areas easily automated and optimized by AI. For non-routine decisions (new business investments, M&A, organizational restructuring, etc.), humans make the final judgment, informed by AI analysis results. This inventory alone will clarify the priorities for AI utilization.
Action 2: Execute a Small-Scale Project
Large-scale AI implementation carries risks. Start by focusing on one department or task and conduct a trial introduction of decision-support AI. Costs can start from a few hundred dollars per month. In my experience, areas like budget vs. actual analysis in the accounting department or prioritizing sales leads in the sales department are relatively easy to implement.
Action 3: Implement AI Literacy Education
Provide AI literacy education to all employees. Deep technical knowledge is unnecessary. The key is understanding “what AI can and cannot do.” Start with a 30-minute study session once a week and have people actually try tools like ChatGPT or Claude. This will quickly reduce resistance.
Conclusion: The Shift to Decision-Making AI Determines Competitiveness
NTT’s AIOWN initiative hints at the arrival of a society where AI is a given at the infrastructure level. The discussion by IT Leaders shows the path to leveraging that infrastructure for management. And the HR survey underscores the need for organizational and personnel change.
What is required of executives is to elevate AI from a “cost-cutting tool” to a “decision-making partner.” To achieve this, three things must be pursued in parallel: building a data foundation, executing small-scale projects, and developing talent.
Check once more if your company’s AI use is stuck at “task support.” If you’re ready to move to the next phase, now is the time.
Comments