The Essence of AI Investment
— Defending Competitive Foundations, Not Chasing Revenue
—
"What AI actually changes for us — that's still unclear."
—
"Without a clear ROI picture, it's hard to justify the decision."
—
"Whether this is truly relevant to our business — I'm not yet convinced."
That instinct is honest. Claims that "AI will grow your revenue" have always felt slightly suspicious. Hesitating to commit to an investment whose impact you can't see is a sound management response.
And yet — I believe that delaying AI investment is itself a management risk.
Not because AI will help you grow. Because choosing not to adopt may itself become a structural risk.
01 —Productivity Gains Won't Grow Your Revenue
AI improves operational efficiency. This is true. Documents get written faster, code becomes easier to produce, research time shrinks. The benefits at the ground level are real.
But the fruits of productivity don't necessarily stay with your company.
Where AI efficiency gains actually go
Industry cost floor drops
Diffuses across the industry
Price competition accelerates
Cost↓ → Price↓ → Margin↓
Competitive advantage retained
Narrowed as tools become common
When competitors adopt the same tools, they gain the same efficiency. The industry cost floor drops, price competition accelerates. This has happened before — when PCs spread, when the internet became ubiquitous, when cloud was normalized. AI is likely to follow the same structure.
02 —The Question Isn't "Use It or Not" — It's "Will a Gap Form?"
AI tools are commodities now. ChatGPT, Copilot, automation tools of every kind — anyone can adopt them, and costs keep falling. The fact that "we're using AI" is no longer a differentiator.
So what creates the difference? The data you run your AI on.
What Is an AI Agent?
An AI agent receives instructions, makes autonomous judgments, and executes work across multiple systems. This is a different dimension from simply asking AI questions. AI takes over the work itself that people were doing. And its "intelligence" is determined not by tool performance, but by the quality and volume of the data it is trained on.
Data-accumulating firms vs. non-adopters — value delivery gap
Whether you start this year determines whether the gap becomes unbridgeable
Organizations that train their AI agents on years of accumulated operational history, decision records, customer-specific know-how, and quality standards — versus those using generic tools as-is — will deliver meaningfully different value within two to three years. That gap cannot be closed retroactively.
03 —The Risks of Delayed AI Adoption
Imagine a competitor begins delivering work of equivalent quality at 30% lower cost, using AI agents. What does your company do?
Non-AI organization
Matching on price erodes margin
Competing on quality is moot — they haven't sacrificed quality either
Competing on speed is futile when agent-powered firms outpace headcount
The existing business model faces sustainability challenges
AI-equipped organization
Delivers equal quality at 30% lower cost
Data accumulation makes agents smarter over time
People are freed to shift toward higher-value work
By the time competitors catch up, the next gap has already formed
This is not a story about revenue growth. It is a story about existing business models facing sustainability challenges.
One Japanese company last year deployed AI to its entire workforce and declared a transition to "an organization where every employee has an AI agent and is expected to leverage it." They didn't add a tool. They redesigned the structure through which value is delivered.
04 —One Critical Step to Take Now
AI investment carries uncertainty. Which technologies will endure, how markets will shift — none of this is fully predictable. This is not a call to go all in. But there is one step that cannot be deferred.
The Compounding Effect of Data — Missing This Year Means Starting From Zero
2025
2026
2027
2028
2029
Current year data
Prior year data
Accumulated data
The data generated through this year's AI activity — the delta between AI-generated drafts and human revisions, logs of business decisions, operational records — cannot be reconstructed after the fact. Data that didn't exist cannot be created retroactively.
The One Thing to Do Now
Start accumulating data. That's all.
No large-scale investment is required. Begin structuring data that already exists within your organization into forms AI can read. Run one small automation, build one proof of execution. Whether you do that this year will determine the gap that appears three years from now.
In Closing —A Decision to Defend, Not to Attack
AI may not immediately grow your revenue. That is true.
But when competitors have strategically deployed AI agents and your organization remains on its traditional footing, competitive capability will erode — quietly, steadily. This is not a story about revenue growth. It is about the increasing difficulty of sustaining the business you already have.
Before "attacking with AI," what's needed is the prior decision: to remain on the side that knows how to use it.
The gap between organizations that have embedded AI agents and those maintaining traditional operations will widen structurally over time.
K.K.
Kazuhiro Kashiwabara
Founder & Representative Director, Advancement Inc.
After careers spanning network engineering, project management and BPR at a consulting firm, and investment execution at a venture capital firm, Kazuhiro founded Advancement Inc. in June 2023. He supports corporate transformation across three domains: AI Transformation, Business Strategy, and Startup Advisory.