Oks argues that the diversified Japanese conglomerate is a coherent adaptation to a financial system built around main banks, keiretsu relationships, and lifetime employment — not a defect to be corrected. Because Japan lacks the hostile takeover market and activist investor pressure that create conglomerate discounts in the US, breadth doesn't get punished the way it does on Wall Street.
The editorial frames the Drucker/Welch focus doctrine as a contingent product of US capital markets — activist investors, quarterly earnings pressure, and a deep M&A market — rather than a timeless management truth. Conglomerate discounts exist because the market can imagine breakups; remove that pressure and the math for diversification changes entirely.
The top comments on the 696-point thread converged on the view that Japanese firms optimize for a different objective function — long-term stability, employment continuity, and keiretsu network strength — rather than shareholder value maximization. Calling them inefficient assumes the American scorecard is the correct one, which begs the question.
David Oks's essay *Why Japanese companies do so many different things* hit 696 points on Hacker News with a thesis that lands harder than its quiet title suggests: the diversified Japanese conglomerate — the kind that makes cameras and life insurance and shipping containers under one roof — isn't a failure of management discipline. It's a coherent response to a financial system that doesn't punish breadth the way American capital markets do.
Oks walks through the canonical examples. Sony sells image sensors to Apple, runs a major film studio, operates a life insurance arm, and ships the PlayStation. Hitachi does power infrastructure, trains, IT services, and construction equipment. Mitsubishi is so many businesses the word "conglomerate" undersells it. In the US, this configuration is treated as a defect — something for activist investors to break up via spinoff. In Japan, it's the default, and it has been for decades.
The HN thread, predictably, became a referendum on Western management orthodoxy. Top comments pointed to the keiretsu system, the role of main banks, lifetime employment norms, and the historical absence of a hostile takeover market. The consensus that emerged wasn't "Japanese firms are inefficient" — it was "Japanese firms optimize for a different objective function, and Westerners keep grading them on the wrong test."
The "focus on your core competency" doctrine — Peter Drucker by way of Jack Welch by way of every MBA syllabus since 1995 — is treated like a law of physics. It isn't. It's a financial artifact of US capital markets, where activist investors, quarterly earnings pressure, and a deep M&A market make conglomerate discounts real and persistent. A diversified American firm trades at a discount to the sum of its parts because the market can imagine breaking it up; that imagination is a price signal, and CEOs respond to it.
Japan's main bank system removes that pressure. When your largest shareholder is a bank that also lends to you, that holds cross-shareholdings with your suppliers, and that has no interest in a hostile takeover, the "unlock shareholder value via spinoff" play stops working. The bank wants you stable across cycles. Stability comes from diversification. The math inverts.
Oks's deeper point — and the one Hacker News mostly missed — is that the doctrine isn't wrong, it's contextual. Focus is the right answer when capital is impatient. Diversification is the right answer when capital is patient and survival across 50-year cycles matters more than 5-year ROIC. Both are rational. Neither is universal.
This cuts against the standard tech-industry critique of Japan — that conglomerates are why Japan "lost" mobile, lost consumer software, lost the AI race. Maybe. But the same structure that produced the Sony Walkman flop also produced the world's dominant image sensor business, which is now critical infrastructure for every iPhone Apple ships. Sony's CMOS sensor division would almost certainly have been spun off or starved under American capital discipline a decade before it became dominant. The conglomerate structure subsidized a side bet long enough for it to become the main business — a pattern American firms structurally can't run.
The counter-case, fairly stated: Japanese conglomerates are also why the country has stagnant productivity growth, why salaries haven't moved in 30 years, and why their software industry is essentially nonexistent at global scale. The patient-capital model preserves stability at the cost of dynamism. You can't have Stripe under a keiretsu. The trade-off is real.
If you're an engineer, this is more than business-school trivia — it tells you something about which organizations can fund the kind of work you actually want to do.
Long-horizon R&D — the kind that takes seven years to ship and three more to monetize — does not happen in companies optimized for next-quarter EPS. It happens in companies whose ownership structure permits it: Japanese conglomerates, family-controlled European industrials, Chinese state-adjacent firms, and a handful of US companies (Google in its founder-controlled era, Berkshire, SpaceX) that have deliberately engineered themselves to ignore quarterly pressure. If you want to do that kind of work, the company's cap table matters more than its tech stack.
The practical implication for technical decision-makers inside diversified firms: stop apologizing for it. The investor-deck pressure to be a "pure-play" anything is a financial argument dressed up as a strategic one. A semiconductor team inside an industrial conglomerate has access to balance-sheet capital that a standalone fab simply cannot match. Use it.
The inverse applies if you're inside a US public company being told to "focus." That instruction is rarely about strategy. It's almost always about the stock price, which is almost always about the next earnings call. Recognize it for what it is, and don't confuse it with engineering wisdom.
The interesting test case is the next decade of AI. Training frontier models requires patient capital on a scale that even the largest US tech firms strain to provide — which is why OpenAI restructured around Microsoft, why Anthropic restructured around Amazon, and why xAI exists inside Musk's broader portfolio rather than as a standalone public company. The structural lesson Japan has been teaching for 40 years — that the optimal corporate form depends on the capital regime, not on management theory — is quietly reasserting itself in the most American of industries. Watch which AI labs survive the next downturn; the answer will track ownership structure more than model quality.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.