LEDE ON
Turns out that September 1 was an important day, as prices changed for many items starting deep in the supply chain all the way up to finished products. Not all these price changes have surfaced worldwide to consumers yet, but they will.
In talking to sources on the semiconductor side, it’s a combination of CapEx (capital expenditures needed for expansion) along with the AI squeeze (the AI needs are sucking up more and more fab time) that are creating the biggest problems for everyone now, even huge companies that dominate some fab use, such as Apple and Qualcomm. I’ll point out that in past chip bubbles—yes, there have been several—the results on the other side are ugly: excess capacity and parts price drops that are challenging to navigate. As the bubble ends, a manufacturer can end up committing to a price only to find that their competitors waited and are paying far lower prices. You don’t just need bean counters these days, you need bean predictors to navigate the semiconductor cycles well, too.
——————————
News
Mirrorless Market Shares
As I continue to dissect the numbers out of Japan for 2025, some things that haven’t yet gotten a lot of coverage: Canon lost market share. Sony lost market share. Nikon lost a tiny bit of market share. OMDS lost market share. Leaving Fujifilm and Panasonic as the only ones making clear market share gains in mirrorless unit volume during the year.
The numbers that were most striking were OMDS’s slip to only 120k mirrorless units in 2025. Fujifilm’s mirrorless sales units went up by more than 2x the total number of units OMDS sold (e.g. 260k unit increase for Fujifilm). That said, Fujifilm remains #4 in mirrorless behind #1 Canon, #2 Sony, and #3 Nikon. Those top three account for 81% of the market. Add in Fujifilm and those four account for 93% of the market (all percentages in this article have been rounded). In essence, they are the market (again mirrorless market). Remember, though, Nikkei/Toyo Keizai really only count Japanese cameras. Leica and Hasselblad don’t show up in their results, for example, but do have enough sales to move the market percentages for the Japanese downward a bit.
If you dig deeper and make a few intelligent assumptions, you also see something else. Overall, Nikon’s share in percent of dollars received is far higher than their unit share (approximately 20% for dollars by my calculations). It’s tough to get a precise accounting of dollar shares for each company here, as only Nikon breaks out enough numbers to do calculations against CIPA and other data. As myself and others have pointed out, Nikon is particularly strong in full frame (FX) body sales compared to the competition, and less reliant upon crop sensor (DX) in sales. The Z5II to Z9 lineup is strong and holding serve.
Fujifilm’s big increase in unit volume does come in the crop sensor (APS-C) market, and that’s exactly where Canon and Sony lost market share last year (based upon my evaluation of some private regional data).
One thing that did surprise me is that implied in all the CIPA/Nikkei/Toyo Keizai reported unit volume statistics for 2025, Fujifilm appears to have sold fewer digital compact cameras than I expected. I would have thought their compact number would have been far higher, particularly given that Fujifilm has a trio of X-players there. But the X100VI keeps going out of stock, so I’m guessing that Fujifilm is not meeting their monthly goals and having to ration 40mp image sensors across their line. The X-T5 probably has a higher gross profit margin, thus that camera may be getting the nod for more of the available chip pile.
Given that mirrorless unit volume is down so far in 2026, no doubt we’ll be seeing a different picture in the 2026 Nikkei/Toyo reports when they appear in mid-2027. Unit volume for mirrorless is down 6% while dollars taken in are up 3%, which is saying that “pursuing volume over price” is transitioning again to “pursuing price over volume.”
Meanwhile, it appears that the Fujifilm-loyalist sites (e.g. Fujirumors) are fighting other sites' (e.g. Petapixel) characterizations of all these numbers (plus making unsupported claims, such as that Nikon has more production capacity than Fujifilm, which I’m not sure is true anymore). It’s unclear to me why who's #3 or #4 are so important. Yes, Fujifilm is growing, and growing faster than the other Japanese makers. Fujirumors makes a point about the three year growth in mirrorless (Fujifilm 97%, Panasonic 93%, Nikon 36%, Sony 18%, and Canon 17%), but we already know so far in 2026 those growth numbers won’t be sustained, as the overall mirrorless unit volume is down 6% year-to-year. With unit volume down, market share growth now has to come at someone’s expense.
Frankly, I don’t care about winners and losers, and certainly not at the #3 and #4 position. I follow this information because it tells me something about priorities. Nikon executives several years ago said that they would emphasize prosumer/pro products, and they appear to have executed to that goal. They’re taking in more dollars a unit than the others are. I’m not sure what Fujifilm’s actual goals are, though. When you press Fujifilm executives they don’t seem to want to talk about that subject, they just want to point to already achieved successes.
What I look at with camera companies is history (how they’ve managed in the past), currency (how they’re managing the current environment, which as you know has tariff, inflation, supply chain and other components adding friction), and what their goals are going forward. Curiously, Nikon executives tend to be pretty good at providing information about the first and third, but not so much on the second (present).
All the current mirrorless products produce great photos. Yes, some have additional performance or features that extend them beyond basic photography use, but what’s been true for some time remains true: if you’re not getting really good images out of your current camera, it’s not the camera that’s at fault. As I predicted some years ago, once mirrorless fully took over from DSLRs, it was going to be tough to keep generating more unit volume with new models and model iterations. Well, we’re there now. The adoption hockey stick growth for mirrorless is now fully traversed, and we’re back to a volume plateau again. Companies are going to have to make splashy moves—i.e. features/performance the others can’t match and that also generate viral user response—to make big unit volume gains now.
——————————
News
Color Me Perplexed
GoPro is merging with Starman, which makes transceivers, and I’m not seeing the connection (pardon the pun). Starman is a nascent <50 employee company making optical interconnections that are designed for hyperscaled data centers. Essentially, those transceivers are designed to provide connectivity between GPUs and servers via optical networks. How that ties into—dang it, ThomAI stop with the puns—action cameras is unclear.
The press release says “creates a unique opportunity.”
Unique indeed. Nicholas Woodman, GoPro’s founder and CEO says this will allow GoPro to “grow across consumer, commercial, and defense markets,” while Starman's CEO cites “GoPro’s world-class optical expertise [sic?].”
Here’s my take: Starman Photonics, part of the Starman holding company behind the merger, is a new, private AI Data Center player. By merging with GoPro they end up on the NASDAQ stock exchange as a publicly held entity and can then try to take advantage of the AI investment surge going on. From GoPro’s standpoint, their debt is completely removed in the acquisition and each shareholder gets about a dollar in cash, though their ownership is diluted 10:1. From Starman’s standpoint, they’re now a public AI company on the NASDAQ and can go around promoting how they’re an undiscovered investment opportunity in the hottest industry of the moment, and oh, by the way, we’re building a US factory. Indeed, the subhead on the press release included “positioning GoPro to expand into AI Data Center, Government, Defense, and Aerospace Markets.”
Charles Tebele, the CEO of Starman Holdings is someone I once met (at COMDEX), and comes out of the personal computer retailing industry, so it’s not difficult to imagine Woodman and Tebele crossing paths at some point, too. Given that Tebele seems to assemble disparate companies under his holding company, maybe this is just another of those transactions. Still, I fail to see the synergy that makes the sum of the parts greater than the sum of the parts.
My doubts aren’t unique. So much negative energy among the GoPro-using crowd was generated by the merger announcement, Woodman felt he had to write a “we’ll keep doing what we’ve been doing” letter to tell the user base that not only will the GoPro lineup continue on, but will be expanded with new offerings now that they’re no longer financially challenged.
I say not-so-fast to Woodman’s accounting here. Looking at GoPro's last financial quarter, the operating loss was US$38.8m on US$104.9m in sales, which represents more than their cash on hand position at the end of the quarter (the merger generates no new cash). The good news is that the new cameras didn’t really show up in that previous quarter, so the operating loss might go away as deliveries really happen. The other good news is that they restart debt-free, so GoPro should be able to borrow if they need more cash.
The problem is that DJI and Insta360 have essentially taken over the action camera market and extended it. GoPro is going to need to move fast and dramatically with product and messaging to get any real traction again. I believe that was the point of Woodman’s letter to customers: he’s trying to stem the tide of folk who are leaking to other brands by reassuring them that they should stick around. The problem is, without details such admonitions don’t tend to do much. “We’ll be great again” is far different than “Here’s how we’re going to be great again.”
That said, the merger has bought Woodman and his GoPro cohort time to show that they can indeed turn their fortunes around. However, given the long decline, the road back up will be tough to navigate.
——————————
News
Keep on DxO-ing
We’re coming into the traditional period where the big software packages get upgraded, and this week we have the release of DxO PhotoLab 10 (hmm, I was just starting to use 9 for a few things). The additions this time are Depth Masks (e.g. foreground, midground, background), a new color control option, Adobe-like People mask breakdowns (eyes, hair, lips, teeth, etc., though this only works for the face), a bunch of UI and performance refinements, and interestingly, the ability to send the image to Affinity for pixel-level changes.
If you’re coming from PhotoLab 8 or 9, the upgrade price is US$130, which effectively means that by “keeping up” PhotoLab 8 users are paying ~US$5/month and 9 users are paying ~US$10/month (see what I did there?). As I’ve long said, the real problem with Adobe’s subscriptions is that if you let them lapse you end up with a shell of the former product, but if you kept them up you weren’t really paying more. At least with the older yearly update model products you can get off the pay wagon and keep using what you’ve got. (I should note that I’m now leaning towards recommending that you don’t do that (get off the paid-for update or subscription wagon), as security issues are rapidly compounding now for those that don’t want to keep up with current builds. That starts at the OS, for sure, but I’ve now started seeing attempts to break into older applications to get access you just don’t want to give malicious characters.)
——————————
News
We’ve Got the Lens, Now About That Body...
Fujifilm this week announced their 400mm f/4.5 R LM OIS WR lens (and once again almost exhausted the alphabet). What’s interesting is how closely Fujifilm mimicked Nikon’s specs with this new lens, as it’s close to rounding errors on most things like size and weight. Also curious is the use of two extremely similar Super ED elements in a key group orienting the light rays for the focus/stabilization point in the lens. This new lens is Fujifilm’s second real “exotic,” and at US$3200 it undercuts Nikon’s excellent 400mm f/4.5 by about US$300.
The problem I have at the moment is that, even with the X-H2s, I’m not sure Fujifilm’s autofocus system is fully up to what I expect when using an exotic (it isn’t with Fujifilm’s earlier exotic, the 500mm f/5.6). Moreover, Sony is about to launch their 400mm f/4.5GM, and undercut both Fujifilm and Nikon in weight, and both the current Nikon and Sony cameras have arguably faster and more reliable autofocus for the situations where you’d use an exotic telephoto. My hope is that Fujifilm has been hard at work on their autofocus system, because right now I find it good but not great, and not at the level of the top three players (Panasonic is even further behind, IMO).
Still, the 400mm f/4.5 is a great new addition for the sports/wildlife crowd using Fujifilm APS-C bodies. A handholdable 600mm equivalent with good light capture attributes extends the usability of their top cameras.
It is interesting that we’re seeing so much activity in the telephoto game at the moment, where the Chinese lens makers aren’t yet really playing.
——————————
News
It Only Took 20 Years
Long-time readers of this site may remember that I started prostelitizing communicating, programmable, modular cameras early in the century. We got one half-hearted attempt at modular in the Ricoh GXR, one half-hearted attempt at programmable with Sony’s PlayMemories system, and everyone’s half-hearted attempts at communications with sometimes-connects-with-camera apps.
German CircuitValley is now doing two of my three (modular and programmable) in a Kickstarter project targeted at researchers, scientists, and tinkerers of all sorts. The CHC5 has modular lens mounts (RF, E, and m4/3 are among those offered) and modular image sensors (various Sony IMX models ranging from 5 to 20.3mp, though the largest image sensor currently available is an m4/3-sized one). A basic interface board supports USB3, HDMI, and Ethernet, but you can also get an interface breakout board if you’re included to roll your own. Software is Open Source, allowing for customization.
The CHC5 isn’t really a mass market product, though, so I’m not going to send you running to the Kickstarter site. I used the word tinkerer earlier, and that’s really where this new camera exists at the moment. However, it illustrates a point that I made 20 years ago: there’s nothing that says that cameras have to be fixed in nature and replaced completely at every upgrade point. It’s a different business proposition than selling one-off boxes for a couple of years and then selling a different one-off box as an upgrade. And it appears that none of the camera companies really want to tackle a different business, even though it would be more environmentally friendly.
As I pointed out when I reviewed the Ricoh GXR and some of its modules when they appeared, the body basics—shape, size, controls, etc.—were relatively locked in by the 1980’s (some might argue for an even earlier date). Interchangeable lenses were an obvious “module” and we’ve had them for well over 50 years now. Things inside the camera kept changing as we moved from analog parts to digital ones, at first gradually, and then dramatically. However, once we got to all-digital innards, it became clear to me that we could modularize that, add programmability and communications and have a whole new beast. Moreover, that would have probably locked you more into a brand than the way the camera makers ended up proceeding.
At a meeting with Nikon executives in 2011, they rejected my ideas for programmability and further modularity, though they ended up running with the communicating thing, which eventually became SnapBridge. While I doubt that moduarity will ever be attempted again in Tokyo and that means that the one-off box parade will continue, I think it’s time to raise the programmability rallying cry again. I’ve been vibe-coding with my local Large Language Module for awhile now, and I can now clearly imagine how you could bring real programmability into the cameras that would allow consumers to make useful functions of their own instead of having to rely upon what Japan thinks they might want.
——————————
Commentary
The World Keeps on Turning
I earlier mentioned that you now pretty much need to stay current with what you’re using (OS, app) from a security standpoint, but Apple this past week also re-iterated that it’s a fully Apple Silicon world moving forward. They notified developers—yes, I’m a registered Apple developer, and an active one—that they can begin dropping universal application support.
A universal application supports both Intel processors as well as Apple Silicon (which is ARM-based). While you can still run Intel code on the impending macOS 27 (Golden Gate) via the Rosetta 2 function, the macOS 28 that will be announced next June will not include that.
Given that photographers are more likely to own a Mac than the general public, I’ve already been counseling a number of them about the fact that they really need to have a plan to move on from their Intel-based Macs. If your Mac is online, it needs to be secured, and the only way to do this is to stay current with the macOS and perhaps add another layer of security, such as Intego’s Essential (antivirus and firewall). You also need a current and updated router, as well, not the router you’ve been using for ten years or more.
As an app developer (surprise!) I have to make decisions about how far back I want anything I produce to support. This isn’t as simple as it might seem. Supporting anything other than the current macOS (Tahoe) and Apple Silicon opens up so many cans of worms that it hurts my brain contemplating it. Since I program in Swift this problem is doubled, as Swift is relative forward looking as a language, not legacy oriented. But the real developer issue is that you probably end up having to virtualize every macOS version you support, plus you need to have virtually every differing Mac architecture available to test on (for current models, that would be Neo, MacBook Air, MacBook Pro, iMac, Mini, and Studio, but that expands if you include Intel-based Macs). For smaller developers, or ones who want to stay nimble, I can see them dropping everything but current macOS support, and sticking to Apple Silicon only. It’s what Apple’s developer tools are optimized for now.
So some advice for photographers:
- If you’re going to stay on Intel Macs — (1) move to macOS Tahoe (if you can); (2) lock in the apps you’re using, as they might not update again; (3) start bolting on extra security for the day you can’t update the macOS again, which is not much more than two years out; (4) seriously consider moving to an Apple Silicon Mac (you’re missing performance, security, and even features from what you’ll be getting from your Intel Mac).
- If you’re on an Apple Silicon Mac — (1) stay current with the macOS updates; (2) make sure the apps you’re using are fully Apple Siliconized; and (3) for those apps that aren’t that you will still require in the future but which have Intel code in them, consider creating a virtualization of the current macOS you can continue to use in the future.
When my new site is ready, I’ll have much more to say about Macs and Mac software. This has been an “investment” year for me, where I’m spending a lot of time making sure I’m ready for the next half decade or so, and so I’ve been dealing with hardware, software, AI, Internet, and a host of other technologies that are in constant movement and flux. I’ve had to update my thinking on virtually everything Mac (and Internet, for that matter), and make sure that I’ve got the right tools for the coming times. At times this has been daunting, as I’m a one-man band playing in an arena with a lot of loud Heavy Metal groups also making noise all around me, but I have to say, it’s been fun. I’m starting to feel a lot like when I was previously most productive, which was the Connectix years in the 90’s. Having a Large Language Model sitting on my desk (literally) is also fun, as I’m using it to accelerate my learning in areas I was falling behind on.
But the point I want to make is this: the constant updating and upgrading on the computer and application side can be annoying, but it’s the cost of staying relevant and productive now. I remember early in my career when all this “digital stuff” started creeping into my analog worlds. That was also annoying as it kept changing rapidly, but keeping up was also one of the things that propelled me to where I am today.
——————————
Commentary
“You Can’t Take a Photograph With a Rumor"
That thought appeared on dpreview recently (@jtr27). It’s something I’ve been preaching for many years now, and links up with the notion that “the best camera is the one you actually have in your hand” thought.
I mention this because some of the rumors that have been making the rounds about some future cameras lately are, well, wrong. Or rather I should write “will turn out to be wrong.” Because I write a lot of things far in advance (sometimes under NDA) and then perform any last minute corrections and editing just prior to posting, I can tell you several upcoming comments I will post when a product releases will have to talk about, at least briefly, those “wrong” rumors.
Rumors are a natural phenomenon, and they do have a mildly useful function in that they tend to surface user-frustrations and expectations in response. We’re getting lots of rumors flourishing in the fora right now because of two things: (1) the lack of any new products to talk about; and (2) the fact that most of the companies have severely tightened how much information gets out prior to launch. The second isn’t just due to the usual attempts to secure the marketing messages, it’s also because those same marketing messages aren’t even known internally until almost the last minute now. I saw that first-hand with the Nikon Z9, where the original internal notion of what would make it into the shipping camera didn’t fully materialize (at least not until three large firmware updates later provided them).
As I’ve noted before, we have a handful of new camera announcements coming this month. Thus, we’ll soon be able to stop talking about some rumors and start discussing the actual product. I look forward to that ;~).
——————————
Commentary
Samsung and Ricoh Need Some I
I noticed in Samsung’s P9 SSD launch that a lot of the materials were marked as being AI generated, though I didn’t report it, as SSDs are a little off-topic for me. It seems I wasn’t the only one to see slop, though. There are now multiple threads and articles from the folk the product was directed at—professional videographers—that Samsung was basically using the type of work to promote their new product that would put those buying the product out of work. Talk about disrepcting your user. Oops.
Cined was pretty direct: “So the drive that is supposed to hold your footage is being sold with footage nobody shot.”
Even the still images in Samsung’s launch campaign were pretty AI-revealing: “talephoto” and “vulture cirsting” were two of the text blocks some image generation AI thought might be relavent to the audience for one photo. That is the equivalent of just slapping the customer in the face with slop that even Hollywood tries to avoid when they do use AI.
Meanwhile, Ricoh appeared to be up to something similar, using AI-generated images to promote a camera that doesn’t take AI-generated images. When confronted, the image was taken down, which sure looks like an admission of guilt to me.
I get it. AI has many useful benefits, and can help someone that knows how to use it correctly to up their game on things like marketing campaigns. But when you simply disrespect your desired customer by slopping nonsense in their face, you should expect that to backfire.
Ironically, the camera companies have been showing this kind of disrespect for decades now, long before AI got useful. Pretty much every year we seemed to get some company using “example photos” for their new camera that weren’t taken by said new camera, but were instead just licensed from somewhere else (and taken with different cameras). Hire a dang photographer and have them actually generate your examples! It’s not difficult. You own the rights. You can show off real capability. You don’t disrespect your intended customer.
Put another way, the camera (and tech) industry needs to be less artificial, and more intelligent.
——————————
Prediction
A Relatively Quiet IBC
IBC (International Broadcasting Convention) starts on September 11th, and as far as I can tell, it’s not going to be a hotbed of new camera introductions as it has been in some previous years. Judging by the early press releases (some under NDA), I’m not seeing a lot of pending news that’s going to make it to this site. More Chinese and Japanese lenses, for sure, but other things in the DSLR/mirrorless space I typically cover seem few and far between. I’m not sure I’d characterize OMDS’s upcoming Pen announcement as IBC-related, though it does come just prior to the show (on the 9th; I’m not disclosing anything secret here, as OMDS has been teasing this release for a week now).
This is not to say there won’t be any newish cameras at the show. GoPro, Sony, and others have had recent launches that will get some of their first public displays and marketing at IBC.
Nikon-owned RED issued a (public) press release this week summarizing what to expect from them at IBC, and that featured areas we haven’t seen Nikon directly involved with in awhile: live broadcasting and immersive 3D experiences (with V-RAPTORs). But in terms of something new, I’m not expecting anything to be announced at the show itself, though it wouldn’t be out of the question to find a new cinema NIKKOR or two behind glass.