<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
<title>Tokyo, but not Tokyo</title>
<link>https://tokyobutnottokyo.com/tech</link>
<atom:link href="https://tokyobutnottokyo.com/feed.xml" rel="self" type="application/rss+xml"/>
<description>Articles on corporate IT, AI transformation, company culture and the price of technology, from the Japan side of multinational companies.</description>
<language>en</language>
<lastBuildDate>Mon, 16 Nov 2026 09:00:00 +0900</lastBuildDate>
<item>
<title>The Blank Box Problem: Why AI Transformation in Japan Needs a Fuller Front Page</title>
<link>https://tokyobutnottokyo.com/article-blank</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-blank</guid>
<pubDate>Mon, 16 Nov 2026 09:00:00 +0900</pubDate>
<description>Most enterprise LLMs open on an empty text box. For non-technical users in Japan, AI transformation needs to look more like Yahoo Japan than like Google.</description>
<content:encoded><![CDATA[<p>There has been a lot of discussion over the years about how Japanese web design philosophy differs from what is popular in the West.</p><p>The Western ideal is sparse. Google’s search page is still a logo, a prompt box and a lot of white space. The design assumes the user knows what they want and asks them to type it.</p><p>Yahoo Japan is the classic counterexample. Its front page carries news, weather, shopping, finance and entertainment at once. Rakuten Ichiba and Kakaku.com are denser still, with product listings, point offers, badges and price comparisons on every screen. To a Western eye this looks cluttered. To many Japanese users it looks complete, because everything they might want to do is visible before they choose.</p><p>Edward Hall’s distinction between low-context and high-context communication is one way to read the difference. In a low-context culture the message is explicit and the listener is not expected to fill gaps, so a page that asks one question and waits for a clear answer makes sense. In a high-context culture, much of the meaning sits in the situation and the relationship. A dense page supplies that situation up front, and the user sees the whole field before deciding where to go. This is my interpretation, and it matches what I have seen working inside Japanese companies.</p><p>I think the same split applies to how companies put AI in front of their employees, and it explains something I watched happen.</p><h2>The blank box at work</h2><p>In 2025, a global consumer goods company I worked for rolled out Microsoft Copilot across its regional operating units. In the Japan and South Korea unit, adoption trailed other regions. The licenses and training materials were the same everywhere, and so was the tool. What differed was the context the tool was being used in.</p><p>Most enterprise LLMs open on a white screen with a text box in the middle and a cursor blinking in it. It is the Google front page, moved into the office. If you are using a chatbot to plan a weekend trip, that works fine. You know what you want to ask, and a mediocre answer costs nothing.</p><p>Work is different. Most people already know the small hesitation of a blank Word document or an empty Excel grid. An enterprise LLM window is the same blank page, with the added problem that it is a new tool and nobody has told you what it is for. An employee has a role, a team, internal rules and a task they may not yet be able to describe in a sentence. Facing an empty box, they have to decide what the tool can do, what they are permitted to put into it, how to phrase the request, and whether the answer can be trusted. In that unit, many people looked at that list of decisions and went back to the tools they already knew. Accounts existed, and most of the activity came from a small group of enthusiasts.</p><h2>What Ax needs in Japan</h2><p>AI transformation (Ax) has to go beyond giving employees a training manual and an enterprise LLM account. For non-technical users in Japan, I think it needs to look more like Yahoo Japan than like Google.</p><p>The first requirement is context. The use cases should be defined and mapped in advance, by department and by task, and placed in front of the user. A sales administrator should be able to see that the tool drafts quotation follow-ups, summarizes meeting notes against the customer record and checks an order form for missing fields. That is the full menu, laid out the way Rakuten lays out a category page. The employee picks from what is visible and does not have to invent a prompt.</p><p>The second requirement is handholding. An employee who is unsure about a new tool will often wait until someone has shown how it fits their work. Walkthroughs, named internal contacts and worked examples built from the company’s own documents carry more weight than a general course on prompting. The training should be attached to the task, so the person learns the tool while doing something they already needed to do.</p><p>The third requirement is purpose-built systems. A general chat window leaves integration to the user. Most of the efficiency gain comes when ML and AI sit inside the systems people already use, such as the expense tool, the ticketing system, the contract template or the approval workflow. The employee clicks a button labeled with a task they recognize, and the model runs behind it. Nobody needs to know what a model is. In a company where decisions pass through formal approval routes, an AI function that is already part of an approved system is also easier to adopt than a separate tool that each person has to justify.</p><p>The fourth requirement is a defined boundary. Employees need to know in plain terms which data they can use, which uses are approved and who to ask when a case is not covered. Ambiguity here is a larger obstacle than any technical limit, and it is a governance question that belongs in the plan from the start.</p><p>Taken together, these are the choices Yahoo Japan and Kakaku.com made years ago for shoppers: show the whole picture first and make the next step obvious. In my framework for Ax, built on the 10-20-70 split between algorithms, data, and people and process, this is the 70 percent. The model is the smaller part of the work. Most of the effort goes into showing the people who will use it what to do.</p>]]></content:encoded>
</item>
<item>
<title>Apple TV 4K Costs Half of My New TV. In Japan, That's a Hard Sell.</title>
<link>https://tokyobutnottokyo.com/article-appletv</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-appletv</guid>
<pubDate>Thu, 22 Oct 2026 09:00:00 +0900</pubDate>
<description>A ¥49,800 box next to a ¥100,000 television, and a smart home lineup built for houses many Japanese households don't live in.</description>
<content:encoded><![CDATA[<p>I recently replaced both of my TVs after 16 years of use. One of them is a 65-inch Regza, and it cost ¥100,000. On October 13, Apple announced the fourth-generation Apple TV 4K at ¥49,800 including tax, or $299 in the US. It goes on sale November 13.</p><p>That is half the price of the TV I just bought, for a box that does what the TV already does. Current smart TVs generally have Netflix, YouTube, Prime Video and the rest built in. The Apple TV has a better interface and a stronger privacy record, and I would pay something for both. Half the cost of a 65-inch television is a lot to pay for them.</p><p>The price is not a Japan markup. Strip out the 10 percent consumption tax and ¥49,800 is about ¥45,300, which comes to roughly $286 at the current rate of about ¥158 to the dollar. That is slightly below the US price of $299, which excludes sales tax. The issue is how large the number is in yen. Most of the climb came earlier. The outgoing 128GB model sold in Japan for ¥23,800 until a June price increase took it to ¥44,800, and the new model adds another ¥5,000 on top of that.</p><p>The HomePod pairing makes it worse. The new Apple TV can drive four HomePod minis or HomePods for a surround setup. HomePod mini 2 is ¥24,800 each, so four cost ¥99,200, and the full setup with the Apple TV comes to ¥149,000. Two speakers for stereo plus the Apple TV already total about ¥99,400, which is the price of my television. An entry-level Denon soundbar costs under ¥30,000 and plugs into the TV I already own.</p><p>The smart home side of the announcement does not fit Japan either, and Apple has announced no Japan availability. The Apple Home Collection, announced the same day, includes a wired video doorbell, cameras, a deadbolt and two thermostats. The doorbell and cameras are listed for the US, Canada and the UK. The lock and thermostats are US and Canada only. HomeView, the new 7-inch smart display, has no announced Japan release.</p><p>The housing numbers show why. In the 2023 Housing and Land Survey, 44.9 percent of dwellings in Japan were apartments, a record. Many of those units have an intercom wired into the building's entrance system, which a tenant usually cannot change. Amazon shows what it takes to serve this market. In April it launched the Ring Doorphone Pro exclusively in Japan, at ¥54,900. It is built to replace an existing intercom using standard Japanese two-wire cabling, and installation requires a licensed electrician. That is a product designed around how Japanese homes are wired, and Apple's wired doorbell has no Japan version. In the US, a homeowner can replace a doorbell in an afternoon.</p><p>Lighting has a similar problem. The standard smart bulb pitch assumes a screw-in socket where the light lives. In many Japanese homes the main light is a ceiling fixture that twists into a hikkake ceiling receptacle (引掛シーリング), most commonly the round type, and replacing a bulb is not an option because the LED is part of the fixture. I have four SwitchBot ceiling lights at home, which are built for that fixture, and they have their own catch. They only respond while the wall switch is on, so anyone who flips it off by habit loses app and voice control until it is switched back on.</p><p>SwitchBot has built its position around these constraints. An estimate by 家電Biz, cited in SwitchBot's own press release, puts it first in smart plugs, remotes, locks and curtain openers in Japan in 2025. Most of those products attach to what a renter already has, without drilling or rewiring. I would expect that kind of product to keep winning in a market where nearly half of homes are apartments.</p><p>Apple's smart home pitch works well for a homeowner with a standard light socket, a doorbell they can replace, and a budget for a second device. Many Japanese households lack at least one of those three. I am not buying an Apple TV for my new television, and I expect most Japanese buyers will make the same call at ¥49,800.</p>]]></content:encoded>
</item>
<item>
<title>Beyond the Balance Sheet: Why Japan's Currency Conversation Needs to Include Tech</title>
<link>https://tokyobutnottokyo.com/article-endpoint-4</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-endpoint-4</guid>
<pubDate>Wed, 14 Oct 2026 09:00:00 +0900</pubDate>
<description>A weaker yen also narrows the way in for individuals who would become Japan's next technical workforce.</description>
<content:encoded><![CDATA[<p>Everything I've written in this series so far has been about cost: refresh budgets, mobile leases, carrier contracts. Underneath the cost story is a bigger one. A weaker yen doesn't just make technology more expensive for corporate procurement teams with negotiating leverage and lease financing behind them. It makes technology harder to reach for individuals, and that's quietly shrinking the pool of technically skilled people corporate Japan will need to hire from in five or ten years.</p><p>The mechanism runs through the same currency exposure covered earlier in this series, just extended further. A memory shortage driving up refresh costs, a weakening yen compounding it, a mobile device and carrier landscape with no real discount tier: all of it traces back to the fact that most of the infrastructure powering Japan's digital economy, hardware, cloud compute, SaaS licensing, and increasingly the AI tools reshaping knowledge work such as Claude, is priced in US dollars. Every one of those budget lines moves in the same direction, for the same underlying reason.</p><p>When the yen weakens, the headlines tend to focus on the export side: record profits at major manufacturers, upgraded earnings guidance, a stock market that rewards a weak currency because it flatters overseas revenue once translated back into yen. That's a real and legitimate part of the story. What gets far less attention is the import side, which is where nearly all of enterprise technology sits. Hardware, cloud infrastructure, and AI tooling are priced globally, mostly in dollars, and are getting steadily more expensive in yen terms with no offsetting upside for the companies paying the bill. The yen has recovered from its summer low, but it remains far weaker than it was a few years ago, and dollar-priced technology has not repriced downward with it. A currency conversation that optimizes for the profitability of a relatively small number of large exporters, with comparatively little visible attention to the cost pressure on the import-dependent side, risks under-weighting exactly the sector Japan needs most to keep growing.</p><p>I got into IT in the early 1990s by building my own computer. A Dell or a Compaq wasn't in the budget, so I built a machine with equivalent capability for a fraction of the retail price, and learned skills along the way that ended up shaping an entire career. That path, spending less than retail to build real capability and learning by doing in the process, is exactly the kind of on-ramp that gets narrower as component and currency pressure pushes prices up. Whatever the modern equivalent of that home-built PC looks like now, it's a harder purchase to justify than it used to be, and that should concern anyone thinking about where Japan's next generation of technical talent comes from.</p><p>The same pressure applies well beyond my own story. A student, a freelancer, or a two-person startup deciding whether to buy a laptop or a phone faces the same price pressure as a large enterprise procurement team, without the negotiating leverage or lease financing that smooths it out for the enterprise. A meaningful share of technical talent and genuine innovation in any market comes from exactly that population, and if the cost of entry keeps climbing, that pipeline narrows. Fewer people casually building technical skill on their own equipment means a smaller pool of technically fluent people entering the workforce over time, at precisely the moment Japan needs more AI, cloud, and data talent. When the domestic pipeline can't keep pace, the practical response for corporations is to look overseas for the skills they can't find at home, adding its own cost and complexity on top of everything else in this series.</p><p>No individual IT organization can fix currency policy or industrial strategy. Those sit well above any single company's budget cycle. But it's worth naming clearly, because it reframes the conversation. This is a competitiveness and talent-pipeline problem that procurement teams cannot absorb quietly inside an annual budget cycle, and it deserves a more prominent place in the broader policy conversation about currency than it currently gets, a conversation that so far has weighted heavily toward the record results of a handful of export champions over the pressure building across the much larger import-dependent side of the economy, including the tech sector everything else increasingly depends on.</p><p>In the meantime, the practical response for IT leaders is the same one I laid out earlier in this series: model FX exposure explicitly across every USD-denominated line item, including hardware, cloud, and AI spend, and start treating currency risk as a first-class input to technology strategy.</p>]]></content:encoded>
</item>
<item>
<title>Your Mobile Lease Didn't Protect You, It Just Delayed the Bill</title>
<link>https://tokyobutnottokyo.com/article-endpoint-3</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-endpoint-3</guid>
<pubDate>Mon, 05 Oct 2026 09:00:00 +0900</pubDate>
<description>Leasing moves price volatility into the renewal number. It does not remove it.</description>
<content:encoded><![CDATA[<p>Most large organizations in Japan moved corporate mobile fleets to leasing years ago, on the reasonable assumption that it would insulate the budget from hardware price swings. It didn't. It just moved the volatility from a one-time purchase decision into a renewal number that's now starting to climb, because the device sitting underneath that lease has gotten meaningfully more expensive to build.</p><p>Apple is now passing memory and component cost increases directly into phone pricing after years of holding US prices relatively steady, and its own leadership has pointed to unprecedented inflation in memory and storage costs as the driver. The iPhone 18 Pro and Pro Max launched in September at $1,199 and $1,299 in the US, each $100 above its predecessor, and at ¥219,800 and ¥239,800 in Japan. The increase didn't stop with the new models. Apple raised the price of every iPhone it kept selling, taking the iPhone 17 from $799 to $899 in the US. In Japan it was the second increase of the year, after one in July, and it lifted the iPhone 17 from ¥142,800 to ¥159,800. Even the cheapest new iPhone Apple sells there, the iPhone 17e, now costs ¥124,800. That second Japan increase arrived after the yen had already recovered from its July low, which fits the pattern that vendor price lists reset on their own schedule and don't follow the exchange rate day to day. There is also no standard iPhone 18 this cycle, so anyone buying before spring 2027 is choosing between a more expensive current-generation phone and a Pro that starts at ¥219,800.</p><p>Google's Pixel line makes the currency mechanism easier to see. The base Pixel model's Japan price climbed from ¥74,800 for the Pixel 5 in 2020 to ¥128,900 for the Pixel 9 in 2024, an increase of roughly 70% in four years, and much of that climb was currency, not hardware. Between the Pixel 7 and Pixel 8 alone, Google's US price rose by only $100, while the Japan price jumped nearly 40%, because the yen moved from around ¥125 to ¥145 per dollar in that same window. The device barely changed. The exchange rate did most of the work.</p><p>The newest generation shows what it looks like when a vendor chooses not to pass that through. The Pixel 11 launched in Japan in August at ¥136,800, 6% above the Pixel 10's launch price, but only because Google dropped the 128GB tier and made 256GB the starting point. Compared at the same 256GB capacity, it is about ¥7,000 cheaper than the Pixel 10 was, while the US price stayed at $899. With Japan's consumption tax stripped out, the yen price works out to an exchange rate of about ¥138 to the dollar, well below where the yen actually trades. Google is absorbing part of the currency gap in a market its regional director called the most important one for Pixel, in what she described as a very difficult year. It hasn't done this across the whole lineup. The Pixel 11 Pro XL costs ¥14,900 more than its predecessor, and the Pixel 10a was raised by ¥12,900 mid-cycle as memory costs climbed.</p><p>In Japan specifically, corporate mobile fleets are overwhelmingly standardized on iPhone or Pixel because of how each manufacturer supports enterprise device management. That leaves IT leaders here with less room to simply shop around for a cheaper device brand the way they might with Windows laptops, where the vendor field is much wider. Within that two-vendor field, though, the gap has opened up: at the same 256GB starting capacity, the base Pixel 11 lists at ¥136,800 against ¥159,800 for the base iPhone 17.</p><p>Here's the part worth saying plainly: a lease price is a function of the underlying device cost, the financing terms, and the residual value assumption. When the device itself gets more expensive to manufacture, that cost shows up in your next renewal quote whether you bought outright or leased. Leasing smooths cash flow and shifts residual-value risk to the leasing company, which are genuinely useful things, but it was never a hedge against the hardware itself becoming more expensive. For a lot of IT budgets, that distinction got lost somewhere along the way.</p><p>Layered on top of this is a structural shift in how Japan's carrier market works. A revision to the Telecommunications Business Act that took effect in 2019 required carriers to stop bundling deep device discounts into service contracts, the practice that used to make the "zero yen phone" possible. Since then, the market has settled into roughly three tiers: full contracts with the major carriers, which come with the fullest set of network features and priority access to newer capabilities as they roll out; carrier-owned budget sub-brands such as ahamo, povo, and LINEMO at a lower monthly cost; and third-party MVNOs at the lowest monthly cost of all. The device price itself, though, is now largely the same across all three tiers, since the discount-cap reform saw to that. The "cheaper" tiers save you money on the service plan, not on the device driving most of the increase in the first place. Feature parity between tiers, particularly newer network capabilities like 5G SA, varies by carrier and is still evolving, so it's worth confirming the current specifics for whichever carriers your fleet uses before treating that gap as fixed.</p><p>A few things follow for anyone managing a mobile fleet here. Lease renewals need to be budgeted as a variable cost, not a flat one. If your forecasting still treats mobile lease costs as steady year over year, that assumption needs updating alongside your laptop refresh budget. The device decision and the plan decision are now genuinely separate line items, since device discounting is off the table industry-wide, and both are worth modeling as moving numbers instead of one bundled figure. It's also worth revisiting device standardization policy. If your MDM and security requirements allow it, check whether the fleet needs to stay this narrowly standardized on flagship-tier iPhone or Pixel models, or whether lower-tier configurations meet the actual need for a meaningful share of users. Enterprise buyers should also push for enterprise-specific carrier terms directly, since corporate bulk agreements are a different conversation from the consumer sub-brand market.</p><p>The pattern running through this whole series shows up here too. A channel that looked "handled" (leasing, or the budget carrier tier) carries the same underlying cost pressure as everything else, just one layer removed from where you'd expect to see it.</p>]]></content:encoded>
</item>
<item>
<title>The Mac Mini Cost Honolulu and Tokyo the Same Labor in 2020. Not Anymore.</title>
<link>https://tokyobutnottokyo.com/article-honolulu</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-honolulu</guid>
<pubDate>Sun, 27 Sep 2026 09:00:00 +0900</pubDate>
<description>In 2020 a Mac Mini cost about the same labor in Honolulu and Tokyo. Six years later Tokyo needs more than twice as many workdays.</description>
<content:encoded><![CDATA[<p>In November 2020, an entry-level worker in Honolulu and one in Tokyo needed almost exactly the same amount of labor to buy a base Mac Mini: 9.0 workdays in Honolulu, 9.9 in Tokyo, less than a day apart. Today, Honolulu needs 7.3 workdays. Tokyo needs 15.3. Two cities that started at parity six years ago are now more than twice as far apart.</p><p>I picked these two cities on purpose, not at random the way most of the pairs in this series have been. Honolulu is where I grew up. Tokyo is where I've spent most of my adult life. They're the two places I actually think of as home, which is part of why watching them start in the same place and pull apart this sharply felt worth writing about on its own.</p><p>The underlying prices: the M1 Mac Mini launched in November 2020 at $699 in the US and ¥80,080 in Japan, tax included. The current M6 starts at $899 and ¥149,800, a 29 percent increase in the US and an 87 percent increase in Japan over the same six years. Apple actually cut the US price to $599 in between, for the M2 and M4 generations, a price Americans held for roughly four years before this year's hikes pushed it back past where it started. Japan never got that dip. Its price climbed the entire time.</p><p>This is exactly why the comparison in this piece uses labor instead of price. Converted to dollars, most of Japan's increase looks like an exchange-rate footnote, the yen weakening rather than something that actually happened to someone. Wages strip that out and ask the question that actually matters: given what a person in that city earns, how much of their own labor does the machine cost now versus six years ago. Using each city's statutory minimum wage, the same benchmark for both years since occupation-specific data doesn't exist going back to 2020, Hawaii's floor rose from $10.10 to $16.00, a 58 percent increase that comfortably outran the Mac Mini's 29 percent US price rise. Tokyo's minimum wage moved from ¥1,013 to ¥1,226, up only 21 percent, nowhere near enough to keep pace with an 87 percent increase in the price of the machine. Honolulu's wage floor moved faster than the computer got more expensive. Tokyo's didn't.</p><p>Currency is part of Japan's 87 percent price increase, but not all of it. The yen weakened by about 44 percent against the dollar over this period, which alone would predict roughly half the jump. The rest, on the order of 25 to 30 percentage points, is Apple's own regional pricing decisions stacked on top of the currency move, and it's a real cost regardless of what the exchange rate does next.</p><p>Apple's positioning of the Mac Mini hasn't moved an inch in this time. It's still marketed as the entry point into the Mac ecosystem, sitting below the iMac and the Mac Studio, aimed at students, developers, and creators getting started rather than professionals with an established budget. That positioning still holds in Honolulu, where 7.3 workdays remains a genuinely entry-level ask. It's a harder claim to sustain in Tokyo, where a computer costing the equivalent of more than two full workweeks of minimum-wage labor isn't really functioning as an entry point for the audience it's supposedly built for, even though the marketing language around it hasn't changed at all.</p><p>Neither city set out to prove anything about currency or wage policy. They just happened to start in the same place and get pulled apart by two different sets of forces moving at two different speeds. Six years turned out to be long enough for that gap to become the story.</p>]]></content:encoded>
</item>
<item>
<title>Apple Held the Watch's Price at $399 for Eight Years. In Japan, It's Up 45 Percent.</title>
<link>https://tokyobutnottokyo.com/article-watch</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-watch</guid>
<pubDate>Tue, 15 Sep 2026 09:00:00 +0900</pubDate>
<description>The US price of the base Apple Watch has not moved in eight years. In Japan it is up about 45 percent.</description>
<content:encoded><![CDATA[<p>The Apple Watch Series 4 launched in September 2018 at $399. The Apple Watch Series 12, announced this month, also starts at $399. Eight years, and Apple didn't move the US price by a single dollar. In Japan, the same product went from about ¥49,500 to ¥71,800 over that same stretch, an increase of roughly 45 percent. Most of that, but not all of it, is the yen. The currency weakened by about 39 percent against the dollar over those same eight years, which on its own explains the bulk of the gap. The remaining piece, a few percentage points on top of that, is Apple's own doing: a margin adjustment beyond what pure currency pass-through would predict.</p><p>I've written before about how Mac mini and iPhone prices have moved in both the US and Japan over the past couple of years, which makes it hard to separate how much of any given gap is a genuine price increase and how much is currency. The base Watch model actually lets us separate those two effects cleanly, because the US price never moved at all. Currency depreciation accounts for roughly 39 of the 45 points. Apple's own margin adjustment accounts for the rest. Nearly all of it happened on the one configuration Apple has never touched in the US.</p><p>The 46mm size tells a smaller, related story, and it's worth including for honesty's sake. That configuration has cost $429 in the US every year since the Series 4 launched in 2018, an eight-year streak Apple just broke with the Series 12, raising it to $449. Japan's 46mm price moved too, by about ¥4,000. Even the one US price Apple did touch this year moved by roughly the same proportion in both currencies. That's really the detail that makes the base-model comparison worth trusting rather than undermining it: Apple does raise US prices when it decides to, and when it does, Japan moves with it. The 42mm case is the one where Apple chose not to, and Japan moved anyway.</p><p>This is the same mechanism I've described elsewhere in this series: Apple periodically resets regional pricing with reference to a dollar-denominated target, and a weaker yen against the dollar over time gets folded into that local price whether or not the US number moves. The Watch is just the version of this story with nothing else muddying it. No memory shortage, no chip generation change driving cost, no US price hike to point to. Just the same object, the same price in dollars, and a currency that's done nearly all of the work on the other side.</p><p>What makes this worth noting isn't really about a watch. It's that a product Apple explicitly protected US customers from repricing on still got meaningfully less affordable for customers in Japan, with no change in the underlying good and no announcement anyone would have noticed. If you're only checking the US price to gauge whether something got more expensive, you're missing exactly this kind of gap, and it's probably showing up in more of your everyday purchases than you'd guess.</p><p>Worth checking the yen price next time you see a US number that looks unchanged. Unchanged in one currency doesn't mean unchanged everywhere.</p>]]></content:encoded>
</item>
<item>
<title>From Problem to Playbook: How IT Leaders in Japan Can Respond to the Endpoint Cost Crunch</title>
<link>https://tokyobutnottokyo.com/article-endpoint-2</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-endpoint-2</guid>
<pubDate>Tue, 15 Sep 2026 09:00:00 +0900</pubDate>
<description>Six levers for managing refresh budgets and device fleets while memory and yen pressures continue.</description>
<content:encoded><![CDATA[<p>I wrote recently about the double cost pressure hitting endpoint refresh cycles in Japan: a global memory and storage shortage pushing hardware prices sharply higher, compounded by a yen that spent much of the summer near 40-year lows against the dollar. The yen has since recovered meaningfully, trading around ¥154 to ¥155 as of this week, its strongest level since February. That recovery hasn't shown up in vendor price lists yet. Regional pricing gets reset periodically, not adjusted in real time, so the yen prices vendors are quoting today still reflect the weaker exchange rate they were calculated against, not the stronger one in effect right now. Until the next price reset, the cost pressure on refresh budgets is exactly what it was when I wrote the first piece.</p><p>So here's a practical playbook. None of this makes the underlying shortage go away, but each of these levers gives you more control over how hard it hits your budget and your fleet.</p><p>First, look at how you acquire hardware in the first place. Negotiating harder on the same ownership model only goes so far. If your organization still buys hardware outright, this is the moment to seriously evaluate leasing or Device-as-a-Service, known as DaaS. Japan's DaaS market is growing fast, projected to reach well over $3 billion in 2026, largely because it converts unpredictable capex into a fixed opex line and shifts residual value and disposal risk to the provider. When component pricing is this volatile, a multi-year fixed-price DaaS or lease contract locks in your cost base while the leasing company absorbs, or hedges, the input volatility.</p><p>Second, right-size specs by role instead of by habit. Many fleets are still running "gold standard" builds set when memory was cheap. With RAM and storage now the most expensive components in any device, a zero-based review of specs by role can meaningfully cut exposure. Not everyone needs 32GB, and not every laptop needs a 1TB drive. This is unglamorous work, but it's the fastest lever you control internally, with no vendor negotiation required.</p><p>Third, move GPU- and memory-hungry workloads off the endpoint entirely. For your specialized users, the ones hit hardest by discontinued high-spec SKUs and multi-month lead times, ask whether the workload needs to live on the device at all. Cloud workstations and virtual desktop options such as Windows 365, AWS WorkSpaces, and Azure Virtual Desktop let you provision GPU and high-memory compute centrally, on demand, without depending on a specific vendor's ability to ship a specific configuration. It also sidesteps the exact shortage making local high-spec hardware hard to source in the first place.</p><p>Fourth, extend your negotiation and allocation timeline. Lead times on high-spec configurations have stretched to months in some cases, and pricing has moved multiple times within a single year. Waiting for the next fiscal year's RFP cycle to start the conversation is no longer a safe default. Engage vendors and distributors early, ask directly about allocation since price alone doesn't guarantee you'll get the units, and consider multi-year framework agreements that lock in both pricing and supply priority, even if it means committing further out than you're used to.</p><p>Fifth, build FX exposure into the IT budget conversation explicitly. Most refresh budgets are still modeled off vendor price lists with no yen-volatility layer on top. Bring finance and treasury into the planning cycle early, and ask whether hedging instruments or yen-denominated vendor contracts are available. Some global vendors will price in local currency if asked, which shifts the FX risk off your budget line entirely. At minimum, model refresh costs across a range of exchange rates instead of a single point estimate.</p><p>Sixth, extend lifecycle instead of defaulting to refresh. Strong asset management pushes a device's useful life further than most standard cycles assume. Proactive maintenance and timely battery or component replacement handle much of it, and extending OS support windows helps too, particularly for frontline roles where obsolescence has usually been driven more by policy than by actual performance need. Certified refurbished inventory is also worth a serious look for lower-tier roles. It's a channel many Japan-based IT organizations have been slow to adopt, and it sidesteps the new-hardware shortage entirely.</p><p>Individually, each of these is a tactical adjustment. Together, they add up to a shift in how the device fleet gets managed: from a fixed annual line item to a portfolio that needs active risk management, the same discipline most IT leaders already apply to vendor contracts and cloud spend. Organizations that make that shift now will be negotiating from a position of control when the next budget cycle comes around.</p><p>What's working for your organization? I'd genuinely like to hear what other IT leaders in Japan are doing to manage this. Comment below or send me a message.</p>]]></content:encoded>
</item>
<item>
<title>The Three Prides of Honda: Corporate Culture in Action</title>
<link>https://tokyobutnottokyo.com/article-honda</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-honda</guid>
<pubDate>Wed, 02 Sep 2026 09:00:00 +0900</pubDate>
<description>Honda's Three Joys read better in English as three prides, and each one gives the person making a decision a different question to ask.</description>
<content:encoded><![CDATA[<p>Today, I want to do a deep dive on a company's corporate culture that actually works, and to reframe it so it translates better to different audiences. Honda's philosophy calls them the Three Joys: 買う喜び, 売る喜び, 創る喜び. The standard translation is "joy" — the joy of buying, the joy of selling, the joy of creating. For an English-speaking, cross-cultural audience, I don't think that's the right word.</p><p>喜び does mean joy. But look at how Honda defines the second one, the joy tied to selling: 誇りと喜び, pride and joy, together. Honda's own text already reaches for pride when it needs to describe what the people building and selling these products actually feel. Joy on its own reads soft in English business writing, close to customer-delight marketing language. Pride carries the weight this philosophy is actually describing, and it travels better across cultures, since pride in your work and pride in what you own are concepts that don't need much translation at all.</p><p>So call them the Three Prides: pride of buying, pride of selling, pride of creating.</p><p>The test of any culture statement is whether it tells an employee what to do when the answer isn't obvious. "Integrity" fails that test. It sounds unarguable and gives no direction, because two people can both act with integrity and reach opposite conclusions. The Three Prides pass it, and they pass it three times over, because each one puts a different question in front of the person making the decision.</p><p>An engineer facing a design choice has a rule: does this produce something I can be proud of creating? Honda defines that joy as exceeding what customers and dealers expect, so the bar isn't "does it meet spec," it's "is this better than what anyone asked for." Soichiro Honda was an engineer, and the company spent decades proving what that standard means in practice through motorsports — motorcycle world championships, Formula 1 engines, programs whose entire purpose was demonstrating Honda could build at the limit of what was technically possible.</p><p>But an engineering culture pointed only at itself never finishes. There's always one more refinement, one more tolerance to tighten. This is where the other two prides do their work, and they don't do it by holding the engineer back. They point the same drive at a fuller target. Pride of selling asks whether the person standing in the showroom can hold their head high offering this to a customer, which is impossible if the product is priced beyond the people it was meant for. Pride of buying asks whether the customer will feel something beyond satisfaction. Both questions are still asking for excellence. They're just asking the engineer to define excellence as something that reaches a person, not something that wins an internal argument. Steve Jobs said it more bluntly: "Real artists ship."</p><p>Three questions, all pushing in the same direction, each one keeping the drive for excellence pointed at something real.</p><p>You can see the result of this on the road rather than on a philosophy page. Cars Honda built in the 1990s and 2000s — the NSX, the Integra Type R, the S2000, the EK9 Civic Type R — still have owners taking pride in them decades later. People restore them, track them, argue about them online, for reasons unrelated to resale value. Honda's engineers pushed naturally aspirated engineering to genuine extremes in that era. Dealers and the enthusiast press could sell that work because the cars backed it up. And the owners' pride has outlasted the depreciation, several model generations, and in some cases Honda's own attention to the segment. That's the full chain running end to end, twenty-five years after the transaction closed.</p><p>Notice what isn't in any of this. Honda's philosophy never says "customer first" or "user-centric." It asks people to build something worth being proud of, sell something worth being proud of, and trust that a customer will feel it. The compass points at the quality of the work, and the customer outcome follows from that rather than being the starting instruction. Which is a genuinely different way to guide a decision than telling employees to do what's best for the customer, since customers want different and contradictory things, and "what's best for the customer" gives an engineer nothing to work with at the moment they have to choose.</p><p>The harder question is whether Honda's current lineup still generates that pride of ownership, or whether the Three Prides have become something written on a corporate site rather than something driving what gets built. That gap — between the culture a company publishes and the culture its actual decisions reveal — is where most culture statements go to die. Honda's has the rare quality of being specific enough that you can check.</p>]]></content:encoded>
</item>
<item>
<title>Why a $799 Computer Doesn't Mean the Same Thing in Every Currency</title>
<link>https://tokyobutnottokyo.com/article-mtai</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-mtai</guid>
<pubDate>Sun, 23 Aug 2026 09:00:00 +0900</pubDate>
<description>A basic computer's price divided by the entry-level wage in each city shows what it costs in labor, not only in currency.</description>
<content:encoded><![CDATA[<p>I built an index to visualize something that's easy to miss when you're only reading prices in one currency: how far a headline price and its real local equivalent can actually drift apart. I read English-language tech blogs and sites regularly, and after years in Japan, I think in both US dollars and yen at the same time. When Apple lists the M4 Mac Mini at $799, it reads as a reasonably priced computer, the kind of price that barely registers as a decision. Look up what the same machine costs in Japan, in yen, and it moves into a different category entirely.</p><p>Before Apple's 2026 price increase, the base M4 Mac Mini was $599 in the US and ¥124,800 in Japan. At the exchange rates through most of that period, ¥124,800 already converted to something closer to $830 to $860. Japan was effectively paying for a machine that looked, in dollar terms, like it belonged in a different product tier, while the US price still read as an entry-level bargain. After the memory-driven price hikes this year, the US price moved to $799 for the equivalent configuration, a $200 jump, while the Japan price moved to ¥134,800, a much smaller proportional increase. The two headline numbers look closer together now. What actually happened is that the US price caught up toward where Japan's price already was.</p><p>The M4 Mac Mini specifically is worth being precise about, since $599 only makes sense as a benchmark once you know what it was buying. Apple positions the Mac Mini as the entry point into its desktop lineup, sitting below the iMac and well below the professional Mac Studio, and has long used it as the cheapest way into the Mac ecosystem. When the M4 model launched in late 2024, it doubled the base RAM to 16GB over the previous generation at the same starting price, and its performance rivaled Macs that had cost considerably more just a few years earlier. Reviewers routinely called it the best value Apple had shipped in years, a genuinely rare description for an Apple product. It's explicitly aimed at developers, students, and creators, not just casual users, since 16GB of memory and Apple Silicon's performance are enough to compile real projects in Xcode or edit 4K video, not just browse the web. That combination, genuinely capable hardware sitting at the bottom of Apple's lineup, is what made $599 worth paying attention to as a benchmark in the first place. It wasn't a stripped-down machine at a loss-leader price. It was a real computer people could use professionally.</p><p>That's the kind of gap a single price comparison hides. A dollar figure reads as affordable or not based on what a dollar means to the person reading it, and most of the currency, tax, and wage context that would make a foreign price legible just isn't there. The index needed to work globally, not just for Japan and the US, and it needed to measure more than currency conversion: what a price actually costs someone in their own labor.</p><p>The approach: take the price of a basic computer in a given city, and divide it by the entry-level wage in that same city. The result is a number of hours, or workdays, of labor it takes to buy the machine. It's the same logic as The Economist's Big Mac Index, inverted: instead of comparing what a fixed product costs across countries, this compares what a fixed institution pays across countries, to measure what a different, globally uniform product actually costs someone.</p><p>Minimum wage turns out to be a poor choice for the wage side. In Bengaluru, for example, using India's statutory minimum wage puts the labor cost of a basic computer at about 126 workdays. Using the entry rate for a junior software developer, a real, formal-sector job, drops that figure to about 50 days. Neither number is wrong, they answer different questions, and a single minimum-wage figure can't hold both. A large informal economy, or a country like the US where minimum wage varies by state, makes the statutory number close to meaningless as a stand-in for what people actually earn. Entry-level fast-food wages work better: they reflect what workers are actually paid, not a legal floor that many employers exceed or simply ignore, and they exist as a comparable job in nearly every country, which is also a deliberate inversion of the same institution the Big Mac Index already uses.</p><p>The Western hubs cluster tightly. London comes in lowest at 5.8 workdays. Berlin and Honolulu land around 6.3 to 6.5. Atlanta, despite being in the same country as Honolulu, comes in at 9.1, a reminder that a single city can't represent an entire country, especially one as large and varied as the US.</p><p>Tokyo and Seoul sit in the middle, at roughly 12 to 13 workdays, nearly double the Western hubs despite both being wealthy, developed economies. Shenzhen requires about 26 workdays, at the edge of moderate access, still well short of the emerging-market hubs but a meaningful step up from Tokyo or Seoul.</p><p>The emerging hubs need the most labor, 50 to 82 workdays depending on the city. Ho Chi Minh City is the highest-barrier hub in this data. Bengaluru and São Paulo, which I initially expected to be the worst-off cities, land closer to the middle of that range once the wage side is sourced consistently rather than estimated per city, a good reminder that the wage source matters as much as the underlying economics.</p><p>None of this is just a curiosity. When a basic computer costs multiple months of entry-level labor, that's not a software problem or an education problem, free coding bootcamps and open courseware don't fix it. It's a hardware problem, and two groups have some ability to move it. Governments can act on the tax side, since several of the largest gaps here are inflated by import duties and VAT stacked on top of the wage gap, so duty relief on entry-level and educational computing hardware would move the number without touching wages at all. Employers can act on the supply side, by routing decommissioned enterprise hardware toward refurbishment and emerging-market hires instead of scrapping it, which lowers the entry price for exactly the talent pool they may eventually want to hire from.</p><p>The next time a price looks reasonable because it's in a currency you're used to, it's worth checking what it actually costs somewhere else. That gap is usually bigger than it looks, and lately, it's getting bigger.</p>]]></content:encoded>
</item>
<item>
<title>Why Most Corporate Culture Statements Don't Actually Guide Anyone</title>
<link>https://tokyobutnottokyo.com/article-culture</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-culture</guid>
<pubDate>Fri, 14 Aug 2026 09:00:00 +0900</pubDate>
<description>Can an employee facing a real decision look at this item and know what to do? Examples from Facebook, Google, Netflix, Honda and Toyota.</description>
<content:encoded><![CDATA[<p>Culture, in the sociological sense, is a set of norms, values, and beliefs that a group of people share and act on. That definition works as well for a company as it does for a country or a religion. The problem is that most corporate culture decks stop at listing the beliefs and never get to the part that actually matters: helping someone make a decision when the answer isn't obvious.</p><p>A good culture item cannot be ambiguous, or if it's stated at a general level, there needs to be a mechanism that makes it concrete. The test is simple. Can an employee facing a real decision look at this item and know what to do? If the answer only becomes clear after a debate about what the words mean, the item has failed at its actual job.</p><p>Take "Integrity," which appears on a large share of corporate values pages. It sounds right, and nobody would argue against it. But it doesn't guide a decision. An employee deciding whether to ship a product with a known minor flaw, or whether to disclose an uncomfortable fact to a customer, gets no direction from the word "integrity" alone, because both choices can be argued as consistent with it. A value that everyone can claim to be following, no matter what they decide, isn't doing any work.</p><p>Compare that to Facebook's early "Move fast and break things." Setting aside whatever you think of the ethics of it, the item works as a piece of corporate culture. It is not ambiguous. It gives an employee a clear answer to a real question: should I ship this code now, even though it isn't fully tested? Culture says speed matters more than polish. If something breaks, fix it after. That's a decision rule, not a mood.</p><p>Google's "Don't be evil" sits on the other side. It sounds similarly punchy, but it hands the decision back to the individual without giving them anything to decide with. What counts as evil is left entirely to personal judgment, and personal judgment varies enormously across a company of thousands of people. It doesn't refer back to a shared standard the group has agreed on. It's not surprising that Google itself moved away from the phrase, quietly replacing it with "Do the right thing" when Alphabet was formed in 2015, which has exactly the same problem in different words.</p><p>Netflix offers another useful pair. The company's actual operating rule, distilled from its culture memo, is to act in Netflix's best interest. A more customer-centric-sounding alternative, something like "do what's best for the customer," sounds nicer, but it doesn't hold up as a decision rule. A customer base of hundreds of millions of people is too diverse and too varied to point in one direction; what's best for a customer who wants more true crime isn't what's best for a customer who wants more anime, and a rule that needs reinterpreting for every individual customer isn't a rule anymore. "What's best for Netflix" works because it has a single, coherent referent an employee can actually reason about. It also tends to land in roughly the same place the customer-first version was reaching for, since protecting the brand, the product, and the service is, in the end, what's best for the customer base as a whole too. It just asks a question an employee can actually answer.</p><p>There's a second layer to this, separate from whether any single item is well written: the difference between explicit and implicit culture. Explicit culture is what shows up in training sessions and on the careers page. Implicit culture is what actually happens: the decisions management makes and the behavior that gets rewarded or quietly punished. When the two are aligned, the culture is strong, because employees see the stated values reflected in real outcomes. When implicit culture dominates and diverges from what's written down, the company's culture is weak, no matter how polished the deck looks. In that situation, the honest move is to not have an explicit culture statement at all. A company that says nothing isn't lying to its employees. A company whose actions contradict its own deck is.</p><p>Netflix's own culture materials make almost exactly this argument. They explicitly call out generic, feel-good values like Integrity as empty unless they're tied to real consequences, and point out that plenty of companies with exactly those words on the lobby wall still collapsed into fraud. Honda has built its philosophy around "The Three Joys," the joy of buying, selling, and creating, reinforced consistently enough across decades that it's genuinely part of how decisions get made, not just a plaque in the lobby. Toyota's culture rests on two pillars, continuous improvement and respect for people, and the company has gone further than most by pairing "respect for people" with a concrete practice, genchi genbutsu, going to the actual place and looking at the actual facts before making a decision. That's the kind of mechanism that turns a general value into something usable. Even Toyota's own brand messaging reflects the same internal direction: "Fun to Drive, Again" wasn't just an ad campaign, it was a public restatement of what the company was optimizing for, at a moment when it would have been easy to let efficiency crowd out everything else. It also wasn't just messaging from the top. Akio Toyoda backed it by example, racing under the pseudonym Morizo in endurance events since 2007, founding what became Gazoo Racing, and personally test-driving development cars to give the feedback that shaped the GR Yaris, the GR Corolla, and the revived Supra and 86. When the person running the company is also the one thrashing prototypes around a racetrack and insisting the results get built, that's implicit culture doing the same work as the slogan.</p><p>None of this means culture decks need clever slogans. It means every item on one needs to survive a simple test: hand it to an employee facing an ambiguous decision, and see whether it actually tells them what to do. If it doesn't, it's not culture. It's decoration.</p>]]></content:encoded>
</item>
<item>
<title>The Endpoint Refresh Cycle Just Got a Lot More Expensive, and Japan Is Feeling It Twice</title>
<link>https://tokyobutnottokyo.com/article-endpoint-1</link>
<guid isPermaLink="true">https://tokyobutnottokyo.com/article-endpoint-1</guid>
<pubDate>Fri, 07 Aug 2026 09:00:00 +0900</pubDate>
<description>A global memory shortage and a weak yen are colliding in Japan. For IT leaders here, they don't just add up. They multiply.</description>
<content:encoded><![CDATA[<p>If you manage endpoints for a Japan-based organization, your next hardware refresh cycle is about to test every assumption in your budget model. Two forces are colliding right now, and for IT leaders here, they don't just add up. They multiply.</p><p>The first is the global memory shortage. AI data center buildouts have absorbed the lion's share of global DRAM and NAND production, and the numbers are stark. Contract prices for standard memory have climbed by well over 50% in a single quarter this year, with some components more than doubling versus 2025 pricing. Analysts don't expect meaningful relief until 2027 or 2028. Every device that ships with memory or storage inherits that cost, whether the vendor absorbs it quietly or passes it straight through in the price list.</p><p>The second is the yen. USD/JPY spent most of the summer sliding toward ¥163 to ¥164, a 40-year low, before Japan and the US carried out a rare joint intervention in early August, the first coordinated yen-buying operation between the two countries since 1998. That pulled the rate back to around ¥157 to ¥158 as of this week, though the yen has already given back close to half the intervention's initial gains, and traders are watching for another round if it drifts back toward ¥160. The pressure behind the slide, wide interest rate differentials and Japan's fiscal position, hasn't gone away, just the acute spike. For organizations here, even the improved rate is still a second markup layered directly on top of the first. Hardware priced globally in dollars gets more expensive to convert into yen at exactly the same moment the dollar price itself is rising.</p><p>For standard frontline endpoints, the laptops and desktops that make up the bulk of most device fleets, this shows up as straightforward sticker shock at renewal or refresh time. Vendors who used to give you room to negotiate on list price now have less room to give, because their own input costs have moved. If your refresh cadence and budget were built on 2024 or 2025 pricing assumptions, they need a hard look before the next cycle, not after the quotes come back.</p><p>For specialized users, the pressure compounds further. Developers and data scientists depend on exactly the components in shortest supply, and so does anyone running AI or ML workloads locally. High RAM and large, fast storage are the obvious ones. GPUs face their own demand crunch on top of that.</p><p>Apple's own lineup is a useful case study. Over the past several months, Apple has quietly pulled the highest-RAM configurations of both the Mac Studio and Mac mini from sale. There's no 256GB or 512GB Mac Studio anymore, and no 32GB or 64GB Mac mini in some configurations, with Apple citing the same memory constraints. Lead times on what remains have stretched to two or three months in some cases. If your organization standardizes on high-spec hardware for specialized roles, the device you speced eighteen months ago may simply not be orderable today, regardless of budget.</p><p>Consumer behavior is already showing what this looks like once it reaches the end user. When Apple raised iPhone prices in Japan by 8% to 11% in mid-July, citing the weak yen, one of the country's largest used-device marketplaces saw order volume more than double within hours of the announcement, a spike its own executives called rare, comparable only to a similar price hike four years earlier. It's worth watching whether the same substitution shows up in the corporate channel: more organizations extending laptop life or buying certified refurbished instead of new. If it does, it will look a lot like this.</p><p>A few practical shifts are worth considering for anyone managing endpoint budgets in Japan right now. Model FX volatility explicitly instead of tracking only USD list price, and build cost scenarios across a range of exchange rates instead of a single point estimate. This is also the moment to find out what rate finance is actually planning to use for next fiscal year's budget. Most multinationals set a single hedged rate at the start of the year and hold procurement to it, which means the cost pressure doesn't arrive smoothly. It shows up as a step change whenever that rate gets reset, and if next year's rate gets locked in anywhere near the recent lows, or drifts back down again once this month's intervention effect fades, the gap between the old hedge and the new one lands all at once instead of gradually. Segment your refresh cadence more deliberately: frontline devices may tolerate an extended cycle of four to five years instead of three, given how much of the current cost increase comes from memory and storage, not CPU-driven obsolescence. Reconsider where specialized compute actually needs to live. For GPU- and memory-hungry workloads, cloud workstations or centralized compute may now work out better than provisioning high-spec local hardware that's back-ordered or unavailable at any price. Engage vendors and distributors earlier than usual, given how much lead times have lengthened, and treat multi-year framework agreements that lock in pricing and allocation as worth the commitment even if it means signing further out than you're used to. It's also worth reassessing "gold standard" specs. If your standard build was set when memory was cheap, ask whether every role actually needs the spec it was assigned.</p><p>This isn't unique to Japan in its origin. The memory shortage is a global, AI-driven supply story. But the yen, even after this month's intervention, means IT organizations here are absorbing a cost that peers in dollar-denominated markets aren't facing at the same scale. Budgets that were already tight going into this cycle are getting tighter, and vendors are already pricing that reality into their quotes.</p><p>Curious how other IT leaders here are adjusting refresh cycles and procurement strategy in response. I'd welcome the conversation.</p>]]></content:encoded>
</item>
</channel>
</rss>
