US vs. Japan IT Industries — "Product" and "Service," and the AI Reshuffle

The same words, “IT company,” point to different things

When you compare the IT industries of the US and Japan, the first thing you trip over is the phrase “IT company” itself. It points to different things on each side.

In the US, “IT company” brings to mind Microsoft, Google, Amazon, Salesforce, NVIDIA, Adobe, ServiceNow, Snowflake, Databricks—companies that take software, cloud, AI, semiconductors, and data platforms to the global market. In Japan, “IT company” tends to bring to mind NTT Data, Fujitsu, NEC, Hitachi, SCSK, TIS, BIPROGY—the SIers and information-service firms that build, run, and maintain systems for corporations and government agencies.

If I had to put the difference in a single line: the American IT industry is closer to a product industry, and the Japanese IT industry is closer to a service industry. Statistically, both are usually classified under the broad umbrella of services. But as industrial structures, their characters split cleanly. The American type reproduces and resells software and cloud platforms it has built once, all over the world—it has the scale economics of manufacturing. The Japanese type builds and operates systems individually, tailored to each customer’s distinct operations, legacy systems, and organizational circumstances—closer in character to construction or contracted business services.

Below, I’ll take this “product versus service” split as the axis and work through the differences in revenue structure and talent evaluation, and how generative AI shakes the whole picture.

America mass-produces “products”; Japan supports “operations”

The core of the American IT industry is turning software, cloud, data, and AI into products and taking them to the world market. Windows, Office, Google Search, AWS, Azure, Salesforce, GitHub, Snowflake—none of these were custom-built for one particular company. They are offered to many customers as a shared foundation.

This model demands enormous upfront investment, but its marginal cost is low. Once you’ve built the software, the cost of serving one more customer is comparatively small, and the more users there are, the easier it is for margins to climb. So American IT companies don’t sell “person-months”—they sell products, platforms, and subscriptions. As of 2025–2026, this structure is shifting even further toward AI infrastructure. The major tech companies are pouring vast sums into data centers, GPUs, power, and cloud platforms for generative AI, and that capital spending draws attention not only as a source of individual competitiveness but as a growth driver for the whole US economy. How it should be reflected in GDP statistics is still debated, but it’s clear that the US IT industry is being reorganized around “cloud + AI infrastructure + software products.”

Japan’s IT industry, by contrast, has long developed around building, running, and maintaining systems for enterprises. Banks, insurers, manufacturing, distribution, telecom, power, railways, government—for customers with huge, complex operations, the main market was constructing systems tailored to their business requirements. Here, every customer’s business flows, data definitions, approval processes, regulations, and legacy systems differ. Rather than adopting a general-purpose product as-is, individual development and large-scale customization tend to win. As a result, in Japan it’s the SIers and information-service firms—not packaged-software companies—that came to hold the greater presence.

Behind this strong “service” character lies the accumulation of legacy systems. Japan’s Ministry of Economy, Trade and Industry (METI) warned of the “2025 Digital Cliff,” where aging, over-complex legacy systems become a major obstacle to DX. JETRO has likewise cited the view that leaving the legacy problem unaddressed could produce economic losses of up to ¥12 trillion a year for Japan as a whole. That’s precisely why Japanese IT demand isn’t just new development; it stretches into legacy renewal, cloud migration, modernization of core systems, data integration, security hardening, and AI-adoption support. IDC expects Japan’s IT-services market to grow at an average of 6.6% a year over 2024–2029, a pace above the global average. Proprietary mainframes, complex bespoke systems, talent shortages, and the pressure to adopt cloud and AI are all pushing that growth up.

As for “Is American IT manufacturing, and Japanese IT services?”—that’s half right and half inaccurate. By classification, software and cloud fall under services too, so strictly speaking American IT companies are services as well. But as a business model, they resemble manufacturing: they design a product, standardize it, deploy it worldwide, and earn revenue by usage or number of contracts. Japanese IT companies get inside the customer’s operations and provide everything from requirements definition to design, development, maintenance, and operation. Less “selling a product” than “providing a service that solves a customer-specific problem.” So the US–Japan difference is more accurately framed not as “manufacturing versus services” but as “a product industry that scales versus a service industry of customer-by-customer bespoke work.”

Change the revenue structure, and the valued talent changes too

This structural difference feeds straight into a difference in what engineers do.

In the American type, engineers build the product itself. Build excellent software, and it reaches millions or tens of millions of users. Competitiveness lives in product design, UX, cloud infrastructure, AI models, data utilization, developer experience, and API ecosystems. So the people who tend to be valued highly are the software engineers, product managers, data scientists, and AI researchers who grow the product.

In the Japanese type, engineers turn the customer’s operations into systems. What matters is understanding the business, organizing requirements, connecting to existing systems, incident response, long-term operation, and coordination across organizations. Japan’s SIers have functioned not as mere coding companies but as “business-design and operations organizations” that keep the infrastructure of corporate activity running. Here the valued people are project managers, business consultants, architects, operations leads, and those who can negotiate with customers.

Under the same job title, “IT engineer,” one side is evaluated on “how far the product it built spreads,” and the other on “how reliably it keeps the customer’s operations running.” The arena is different.

Generative AI both widens and narrows the gap

Bring generative AI into this, and the US–Japan gap could either widen or, conversely, narrow.

In the US, generative AI is developing as an extension of the existing cloud and software industry. Large models, AI agents, coding assistance, search, advertising, business apps, data analytics—one after another, they’re embedded into platforms that already exist. In other words, AI works to further strengthen the American “product industry.” Put AI on a standardized product and ship it to the world—their strong suit works exactly as before.

In Japan, generative AI is starting to be used for operational efficiency, internal-knowledge utilization, inquiry handling, document creation, development support, and legacy renewal. The market’s growth is large, too: JETRO cites a forecast that Japan’s AI market will expand to about ¥1.7774 trillion by 2030. But what’s really being asked of Japan is not only building the AI models themselves. Rather, it’s the ability to embed AI safely into existing operations, core systems, on-site data, manufacturing equipment, financial operations, and administrative procedures. Here, the business understanding and systems-integration experience that SIers have carried for years can become a strength as-is.

Japan’s weaknesses and strengths, and the direction of its evolution

Japan’s weakness is clear. It’s weak at turning software into products and scaling them in the global market. Build an excellent bespoke system, and it doesn’t spread easily to other companies or overseas. Too much customer-by-customer customization makes software assets hard to reuse. On top of that, the multi-layered subcontracting structure and person-month contracts tend to hinder investment in products and the accumulation of technology.

The strength is just as clear. It is closeness to industrial frontlines. Manufacturing, logistics, healthcare, finance, public services, transport, energy—Japan has many IT firms that deeply understand complex on-site operations. In the age of AI, simply owning a model does not create value. Far more important is being able to answer: how do you handle on-site data, which business decisions do you put AI into, where does a human check, and how do you shut things down safely when something breaks?

In this sense, Japan’s IT industry has room to evolve from a “contract-development industry” into a “business-transformation industry for the AI age.” The key is whether it can convert the knowledge that has been locked inside individual projects into reusable products, templates, business platforms, and AI-agent foundations. Whether it can cross that line is the dividing point.

The future US–Japan gap will no longer be a simple story of “America strong, Japan weak.” America is overwhelmingly strong in AI models, cloud, semiconductors, and software products. But its AI-infrastructure investment has grown so large that concerns are mounting over monetization, power constraints, and payback. Japan lags far behind in foundation models and cloud, yet in legacy renewal, operational AI adoption, industrial DX, the softwareization of manufacturing, physical AI, robotics, and mobility DX, it can hold competitive territory of its own. In its mobility-DX strategy, METI has set a target of a 30% global sales share for software-defined vehicles (SDVs) in 2030 and 2035. It’s one sign that the boundary between the auto industry and the IT industry is dissolving.

Summary

• The difference can’t be fully explained by the simple label “America is manufacturing, Japan is services.” More accurately, America is a “product/platform type” that deploys products to the world, and Japan is a “service-integration type” that supports customer-specific operations and existing systems.
• That difference has produced gaps in revenue structure, talent evaluation, technology investment, corporate culture, and international competitiveness.
• Generative AI further strengthens the American product industry, while it asks of Japan the power to embed AI safely into the frontline, the operations, and the core systems.
• Japan’s weakness is productization and scale; its strength is closeness to the frontline. The decisive point is whether it can turn the operational knowledge earned in individual projects into reusable software assets.

Whether Japan can accumulate the insights gained from customer-by-customer work as AI agents, business templates, data foundations, and industry-specific platforms—that, I think, will be the single biggest key to narrowing the US–Japan gap from here.