227. CES 2025: An Abundance of (AI) Experimentation

CES proved a transition year with many established product lines showing incremental improvements or focusing on B2B areas and new product lines showing a rush to demonstrate their relevancy to AI.

This year’s enormous Consumer Electronics Show proved to be a transition year with many established product lines showing incremental improvements or focusing on B2B areas and new product lines showing a rush to demonstrate their relevancy to AI. It’s still early.

Let’s get two things out of the way.

First, CES continues to be HUGE. I mean literally the show is just enormous and has sprung back to life post-pandemic. The official numbers were over 140,000 attendees (from over 160 countries) and 4,500 exhibitors (inc 1,500 startups). This year the show spanned the entire Las Vegas Convention Center and Venetian/Sands Expo Center plus countless hotel suites and meeting rooms. I walked 83,000 or so steps which wasn’t a record because stuff was a bit more compressed this year with the LVCC fully open.

Second, despite the name Consumer Electronics Show, the show is not for consumers and more than ever before I feel the show has toned down the idea of direct-to-consumer launches, communications, and booth content. I do think that is a positive, but it also makes the show a bit more difficult to describe and to make seem exciting. In past recent years I have referred to this as “ingredients” or “platforms”, but I think in many ways the shift is complete. It was the iPhone, which is famously not directly represented by Apple at the show and perhaps before that the switch to Yahoo/Google/Facebook as exciting, that began this trend away from CES as a shopping trip to CES as a true deep industry show. Again, this is a net positive. Probably the most broad-based consumer product at the show remains TV sets even with mobile phones in use in every booth and laptops still making news. TV has a special place in the origins of CES. Even home audio has seen a decline and with the “home theater” becoming a far more specialized niche that has seen a decrease in booth space as it has in retail showrooms. While Auto is a huge industry and direct to consumer it always felt a bit out of place at CES and I am glad to see this become more of the technology than “cars” even though that is still a bit weird.

A favorite thing that I see every year are attendees literally shopping and learning that you can’t actually buy anything, and most cool things are non-working prototypes intended to gauge or generate interest in jobbers. For example, there was a super cute 3-in-1 magsafe Apple product charger shaped like a Cybertruck that I watched a half dozen people try to get one unit. The rep handled it all well offering a great price on thousands of units available in 12 weeks. Back in the old says when products were super heavy you could often pay cash and grab a sample at the last days of the show just so they didn’t need to ship things back. Not anymore.

The biggest structural changes in the show include the following. The presence of the largest tech companies including Google, Meta, Microsoft, is minimal to none really. This year saw a keynote from Nvidia (following past keynotes from AMD and Intel for many years) which is detailed below. The largest booths continue to be the Korean consumer electronics companies, Samsung and LG. The largest Japan booth is Sony but even it is a shell of what it used to be. Thankfully the US and German auto makers have toned down their booths which took up a lot of space and contributed little information, replaced with a wide range of booths covering component (some startups) for autonomy and EV (I won’t cover these here except for brief mentions).

The largest structural change continues to be the combination of country-sponsored booths, and the Eureka Park booths focused on startups. The country booths were historically a bit of a gambit to get the various trade and business sectors of governments to foot some of the bill for renting the space by aggregating all the companies from one country. This worked when it was showing off everything from Austria or Poland. But now nearly every country has a booth which makes understanding what is going on difficult. As an example, nearly every booth from Europe has healthcare tech to show off, much sponsored by the national government. But if you’re trying to get a sense for healthcare in general you find yourself navigating a sea of country booths and context switching between healthcare, smart cities, and AI translation apps for one country and then another. Additionally, some of the country booths are enormous. The mainland China booth occupied half of a large hall with everything from literally USB connector components and neodymium magnets all the way to autonomous taxi drones. At this point the show would be better if CES could solve the booth rental economics while keeping the themed products and companies together. The Eureka booths are also organized some by country and some by themes, but that area is so compact simply walking the booths systematically is exhausting.

A quick word about the world we live in. Prior to the show there was the attack on the Trump building. This amped up security relative to initial plans. It was clear there were some last-minute changes and thus not everything was smooth. Much like post 9/11 COMDEX this was all necessary and warranted but could be frustrating (the Nvidia keynote was quite late to start for example). Additionally, heavy storms over much of the US created havoc with those flying in from the east which includes much of Europe. Attendance on day 1 was thinner as a result, which was great if you made it there. Then on Tuesday (first full day of show) the LA fires started. Suffice it to say many many attendees and exhibiting companies were directly impacted by these. When you bring 140,000 people from around the world physically to one place, a lot can happen. A lot happened this year. 🙏

Experimentation

The major theme was experimentation specifically with AI. What I mean by this is that much of the emphasis on products being shown was showing as much as they could credibly get done between a given company’s recognition of the enormity of AI (specifically GPT/LLM) and the show. In SV a year ago the phrase was “AI wrapper” and many companies have moved on from that view. In the world at large I think we’re still seeing a lot of AI wrappers. This is not a negative and is not only expected but it is necessary. This is literally how innovation happens. The first ideas aren’t always the best, but you have to traverse this idea maze to get to the good ideas.

I can’t stand reviews of the show that talk about something not being useful right away or “trivial” because everything we’re seeing will be seen by someone who will take it to the next level. Even if it isn’t the company seeing it the engineers will move on, and the product people will be influenced. Today when you see something like SharePoint of Notion, one has to appreciate that in the late 1990s the show(s) were filled with collaborate “intranet web sites” that often did what look exactly like these products today. Not only did I see those products but I wrote up my trip report and can draw a straight line of influence to what came to exist (don’t believe me, go read about eRoom as one of a dozen examples).

Themes

I will break down the rest of the report into the following themes:

• Nvidia Keynote

• Al

• Health Wearables

• Glasses/Headphones/ Smart home / Home

• Auto / Transport / Drones

• TVs

• PC/ Gaming / Laptops

• Smartphones and

• Robots

• Gadgets

I will include photos to show off some neat or interesting demonstrations. There’s no attempt to show off everything or endorse one specific. Often it was just which demo I was able to get a better photo.

But here are 7 BIG things to think about:

• AI is the new ingredient. But right now, no one is sure what to do with it other than add it to their existing products and that feels awkward though for some products it means they finally work as expected, but will that change things?

• Screens are everywhere. But right now, there are just too many screens. In a world where people are complaining about screen time, I just don’t know if adding a screen to an oven or a screen to my phone charger make sense. As with the graphical interface itself, there were times when using a GUI just didn’t make sense even though it was the thing to use. My feeling is few will find adding a stationary small screen for either home control or entertainment compelling, especially if it drags in another ecosystem. For better or worse people walk around their house with their phones and even if they don’t the reaction will always be to grab their phone. Plus, the security and privacy implications are real.

• Health telemetry is here, economical, and far more interesting. Finally. The thing the health telemetry was missing was the ability to seamlessly and continuously in the background measure things non-invasively. Some of us are seeing the benefits with sleep that can come from the right device with the most seamless access. Apple Watch especially with vitals is showing how a small number of regularly measured points can be quite useful. CES 2025 showed we’re on the verge of being able to measure blood pressure and glucose in addition to new measures in ECG, respiration, sleep apnea, and more. These are real data points and when combined with relative change versus precision measurement (“medical grade”) we can see the way health is substantially making progress. I have been kind of skeptical in the past, but innovations are critical mass now.

• Supply chain is adapting to the world. The supply chain is starting to change in response to geopolitical. For the first time I saw China booths talking openly about factories outside of mainland. I think this is a really big deal if you make things. There is a long way to go, and the end point is unclear. But starting with small things like phone chargers is interesting in a big way.

• Time from ODM to product. CES used to be about products you could see in a prototype stage one year that would be available in a year. Now the ODM (the factory) is prototyping and showing proof of concept while at the same time they’ve already started selling early versions to Temu and other nameless OEMs you can see on Amazon or Instagram ads. I had to really struggle to find anything that I had not yet been pushed an ad for on those platforms. Kind of incredible.

• Home appliances, security, automation have made a ton of progress over the past few years but now seem to have plateaued. The rise of AI might prove more of a distraction when we still have a ton of last mile work.

• Data and privacy are at a critical mass of invasion. Cameras, locator beacons, every device phoning home, and more are at a critical point in terms of privacy. CES shows the huge interest from governments in face recognition, cameras, and location data. Ironically the self-proclaimed guardians of privacy in the EU give governments themselves essentially a free pass on these issues. I personally worry deeply about a) presumed guilt and the role all this data plays in exposing innocent people to get caught up in the system, and b) the way the court system works and how everything becomes evidence. I won’t have more to say in this report, but I could not help but see almost every device and sensor through the lens of the loss of domain and the presumption of guilt.

What do I mean by this last one? Every year there is some booth that just triggers me with respect to privacy. Here was this year’s. Talk about dystopian.

Just a reminder, I’m just one person walking the floor. I don’t see everything no matter how systematic I might be trying to. In particular, I can’t compete with a web site that dispatches a dozen reporters and videographers. Just me. And 35 or so years of going.

Also, I don’t include links because I don’t want anyone to think I’m endorsing or monetizing this in any way. Do your own research ;-)

Nvidia Keynote

To say the Nvidia keynote was the big event would understate things. This is super weird for me to say since I was there when Bill Gates keynoted the conference and even announced Xbox (along with Tablet PC and Windows 2000). Microsoft was also the largest market cap company at the time (and also under intense antitrust scrutiny) and the darling of Wall Street. While waiting in line at the event to enter I even saw long-time CTA CEO Gary Shapiro (owners of the show) and had a quick word about this compared to those old keynotes before he showed me the shortcut to get in. The excitement, energy, and scale of this keynote dwarfed anything I could recall. It was at a scale we saw for some Windows events in China where getting large crowds wasn’t ever a problem. Estimates were 10-12,000 people at the keynote held in the Mandalay Bay theater. It was pure insanity to think this was happening because of…graphics cards ;-)

Jensen was Jensen and brought with him a depth of technical knowledge, industry experience, and founder credibility that you just have to love and experience. He is the leading tech CEO right now in terms of innovation, execution, and excitement across hardware, software, consumer and enterprise. Nothing compares.

There were several important announcements in his presentation which was just him on stage talking and demonstrating. No net.

By far the biggest announce was NVIDIA Project DIGITS. DIGITS builds on the immediate product announcement of the Blackwell net generation chipset/software for AI and graphics (below). With the GB10 Grace Blackwell Superchip, Project DIGITS delivers a petaflop of AI performance in a power-efficient, compact form factor. It is basically a Mac Mini formfactor AI supercomputer. The compute platform consists of the Blackwell chipset along with a CPU provided by MediaTek. Here’s a shot from the web site which is clearer than what I could capture showing it off:

The key with this device is that you can train and infer models on one device at some scale. It is impressive. It runs a Linux OS (DGX OS) with Nvidia CUDA baked in running with 128GB of unified, coherent memory and up to 4TB of NVMe storage. It is a beast in a tiny package. The idea is for this to be the dev platform which then can seamlessly scale to Nvidia DGX Cloud instances. By the buzz walking out this is what everyone wanted NOW. Gamers were wondering if you could run Windows on it :-;

That’s the interesting part to me of course. It is entirely accessible to Windows/Mac developers via a remoting protocol or as a server. The processor it is running is from MediaTek. The Nvidia GB10 brings NVIDIA Grace CPU, which includes 20 power-efficient cores built with the Arm architecture combined with Blackwell. Jensen showed off a wafer to illustrate how much is jammed into the chip. There’s a massive amount of manufacturing and design prowess to deliver all this reliably.

My sense is this was the first step in Nvidia delving back into the CPU product line in a general-purpose platform (v autos). I have no insider roadmap information at all but can say we made a bet for the original ARM Surface for Windows on Nvidia precisely because the lead in GPUs mattered so much. We tried to create a new class of device when we announced that in 2011 CES (that’s how Jensen described it then). The industry would be greatly served by Nvidia offering the combination of technologies that could be tapped for general purpose compute. Qualcomm is not even close on the compute that matters for AI IMO. It might be that the consumer will benefit indirectly if this platform is just used by developers and data centers, but my strongly held view is that the devices we carry will end up doing much of the work we think of today as hyperscalers or cloud, so we need this to happen.

The exciting news for Blackwell was really about software and it shows the enormous lead Nvidia has as both a hardware and software platform. The Blackwell chipset of course brings a massive amount of compute power, but much of the benefit happens through the use of execution of AI models for graphics rendering. The use of AI dramatically reduces the real time compute needs and replaces those with generated/inferred frames and pixels. It essentially takes shading and rendering to a new level of software generation, a new level of abstraction.

The result is that Blackwell is $1000 cheaper! And there is a laptop version that was announced with partnership of many OEMs.

Nvidia under Jensen’s leadership for two decades has an incredible ability to invest for the long term, try new things, and take risks. For those in the software world it is sometimes easy to forget how the hardware world takes 3 years of investing to see new products. It brings a whole new meaning to agile management and Jensen is the very best at this.

The DIGITS announcement is the biggest and most strategically interesting thing I believe we will see in 2025. It is too easy to look at it as a simply a supercomputer. CES can make people do that. But here’s what I think makes is strategically the most interesting of announcements:

• Whitespace. The product is industry whitespace. The industry is focused on the edge with some custom silicon like Apple’s NPU (very capable), but no one is doing what amounts to a full AI-capable computer on “the desktop” this way. This has huge implications for development, particularly cost of development.

• APIs. The key to this development is a consistent set of APIs on desktop and at hyperscale. While many models can run locally today (the open-source ones for example) the unique ability for developers to tap into the platform at these different scale points can solidify the Nvidia software stack.

• Models. Underlying this Nvidia is taking steps to make sure the open-source models work effectively with their APIs so the silicon optimizations shine. Historically, hardware makers have embraced open source in order to tap into their proprietary hardware when the community in general does not do that. In this case, Nvidia has a huge decade’s interest and support for its evolving platform which makes this more than a wish but a strategic point.

• CPU. While the CPU and general-purpose nature of this version one “project” were limited, it continues to push ARM at both the edge device and the cloud level. I don’t think this is the last we will hear about Nvidia CPUs.

• The hyperscalers are focused on cloud consumption. All the startups are feeling the financial aspects of this right now. And as consumers the idea of subscribing will soon show itself to be kind of a crazy waystop as any AI apps we use will have to build in the very same subscription (Apple worked around this with the pre-signin and jump to GPT). At the enterprise level, it seems highly unlikely that an enterprise with 50,000 employees can utilize a variable cost SaaS product to streamline anything when doing so means unbounded/unbudgetable costs. Without writing a whole essay on this, it all feels very unstable much like the early days of the centralized scaled Cable TV companies or early smartphones and the carriers thinking they could charge everyone for everything on the “world wide web” as gatekeepers.

If you’re excited about AI and how it evolves and (like me) also think the current foundation model/chat hype cycle is seeing its limits, then I think what Nvidia showed off is the precursor to exciting things to come from developers who tap into the architecture enabled by these solutions.

This was just so fantastic to see.

AI

Everyone at CES got the memo. AI is the most important thing. This was like CES 1998 when everyone got the memo “Windows and the Internet”. If there hadn’t been a dramatic move away from swag and paper/cardboard, then there would have been “AI HERE” table toppers and stickers all over the show. This is important because not only is AI important and has the potential to deliver a new wave of innovation, but there was also a need to do new things. Much of the industry has been caught in a “what comes next” lull and with the exception of the ongoing iteration in healthcare (below) and the bright spots in robots and XR we haven’t seen much and definitely have not seen much with any broad deployment, perhaps except Apple Watch for health and some other health devices seeing moderate take-up in the larger/wealthy markets.

So, there is much optimism and excitement.

However, that optimism and excitement was met with mostly just marketing. Try as I might, I just didn’t find any product concepts/experiments that were really about AI and all that interesting. The way I came to think about this was as follows.

Imagine if available to everyone were the following. Your product could, with almost no work:

• Convert any spoken language into text

• Convert any text into speech

• Convert any handwriting into text

• Translate between any languages

• Recognize any object in a scene, including a specific person

• Describe any scene

• Answer any question with a well written (or spoken) answer

• Create a video persona doing any of the above

There might be some more, but you get the point. The models today can do all those things. So, what we were able to see on the show floor were all the existing products one could imagine (and have seen previously) “hooking up” these capabilities.

For example, CES has always had translator apps. These go way back to the origins of CES and attendees needing to talk to each other. Maybe the companies are different today, but the scenario is the same and the products are vastly improved. At the same time, everyone is just using their phone and widely available apps or even native foundation models.

CES has had eInk notetaking devices for ages. eInk is a perennial tech waiting to break out in scale. Kindle finally got traction but is still not a mass scale device. Color eInk is finally hear and can be used for readers and other novel products like low power digital signage. Now the readers can have written ink and with the models this ink can be recognized and turned into text. Personally, this is incredibly exciting because the original Microsoft research was founded with a handwriting recognition lab…in 1992 or so :-; But again, this is just the same scenario only now it mostly works.

Security cameras have been a mainstay at the show since China really started attending in force. The cameras moved from analog to digital, from wired to Wi-Fi, from daytime to night, and so on. Now they all have object recognition and scene description. There are many fancy features such as recognizing different animals or detecting anomalous scenes and being able to alert the operator directly. Still, these are security cameras, but AI enabled. Just need to hope the summary and ID are accurate (see photo).

Below is an AI assistant that is a generated persona that interacts via speech and voice. It can translate and so all sorts of stuff by stitching together what are all the pieces available to everyone today. The question is really the last mile.

The thing about where we are with AI is that we’re two+ years post the big rise of chat. The main problem chat needed to solve that went beyond what other models solved (translation, recognition) was to be truthful. Stitching all the pieces together is not solving this core technology problem. Of course, this can be and will be addressed but to do so is likely going to revisit much of what made chat great (the creative side) and likely will diminish the role of the foundation models. In other words, what is so important today will be far less important parts of the eventual products. This upends the tech stack and the leaders in the process.

Pushing against this is the relentless cycle of what comes next. From my perspective, we did not come to address truth but have leaped to “AGI is now solvable”. That’s a big leap since presumably AGI requires some level of truth. And many have taken to proclaiming the agentive future. But this skips past both truth and AGI. The vocabulary and expectations are racing far ahead of what is being delivered.

Returning to the above AI assistant. I have 0% doubt it can address basic Q&A say at a theme park or a mall, assuming scoping of the results. The problem is the rush to think this solves all sorts of service problems for the enterprise. As we all know, at least anyone who has written line of business software, the problem is never the known case but the exceptions. The models and frankly AI in general are not equipped to handle exceptions. So, for me the leap to agents and autonomous “do this work for me” seems like a promise that can’t be kept. This makes me uncomfortable. My challenge will be that the demos will be great. But I suspect that the inputs to agents are very quickly going to look like programming more than like asking a trained human for help. Or worse it will be the voice response version of a call agent who is not empowered to make things right but to only follow the flow chart.

It is important not to read this as negative. It is a state of where we are. I get how transitions happen and more than anything I understand what it is like to have negative vibes shouted from the cheap seats. What I am looking forward to more than anything are new products that use AI to do things that actually work better. The web disintermediated a vast array of physical and human barriers to getting to information, products, and services. That’s what is going to happen but what is disintermediated, how, and when is still not obvious. It is obvious to people taking risks and doing experiments. Those aren’t always the booths at CES.

The flip side of all this is the over-the-top AI marketing. The major CE companies all got the memo. They put AI in front of everything!

Note, these next sections will be much briefer because of the overwhelming import of AI and the amount I covered it.

Health/Wearables

After years of skeptical coverage of health and wearables, I am feeling more optimistic than ever. I give Apple a ton of credit for moving slowly but deliberately in the space as that has created a foundation upon which the ecosystem can work. If you’re into health metrics and are bought into the Apple ecosystem, the Health app and third-party ecosystem really does offer a “whole new world” even if US people include Epic.

There was a good deal AI marketing. The biggest example was the Withings “body scanner” which was really the existing Withings scale (latest body composition model) with a full body scanner mirror offering an AI analysis and a televisit. But really it was a prototype leveraging their existing and quite good scales and other devices. While they offer a service (and subscription) they also integrate with Apple Health which to me shows the way towards progress while also building on trust and privacy.

Blood Glucose monitors are a perennial showing (going way back) and have just struggled to find a non-invasive measurement. Many reading this have finally been able to tap into continuous glucose monitors (CGM) with Dexcom through Levels or a doctor that are now OTC. These have pros and cons. Finger stick is still the gold standard but for building body telemetry is problematic. Most of the innovation is happening outside the US (Japan and Korea) where the regulatory framework is more permissive. Ortiv is a new kind of fingerstick approach that uses a laser to get the sample. Many continue to work on using various wavelengths of light and laser to measure—these require AI to analyze the reflections on the sensor. There was a near-field IR this year that might be promising but might not fit in a watch.

This is another non-invasive glucose monitor from a company based in Seattle. It uses nearfield IR. I watched it work. So, I am hopeful.

This was a cool product I loved. It is a GPS tracker built into insoles. It is aimed at either the elderly or kids. The benefit here is that it isn’t highly visible and stays out of the way while kids are playing. In general, the AirTag problem is that any bad actors know to remove them from people or objects, so we need more way to disguise those.

Like glucose levels, measuring blood pressure has long been without a continuous mobile solution. Many things have been tried but so far, no luck. Novosound is a UK company specializing in ultrasound. They have developed an ultrasound sensor that can be part of a watch/band and measure BP. I spoke with a member of the team building the ultrasound part (versus the full device) who seemed particularly optimistic that it can be deployed in a consumer-friendly approach. Having continuous blood pressure would be a huge breakthrough for both healthy and others. There’s very little data on how BP various over a normal day and having data for an individual relative to day after day would be HUGE. Below is the ultrasound probe that could go in a band (in her hand is a stand-alone prototype with battery).

If you have not gotten a walking pad for home, then you’re missing out. If you had a treadmill and couldn’t move it out of your basement or up the stairs then get a walking pad! These are amazing for those of us in a rainy city. They are incredibly cheap and easily carried. Amazon is filled with them as is my Instagram feed.

While much more about AI than health, one area that many are super bullish on is AI reading any sort of medical images. Living with a person who reads images for a living I am less bullish on this area. The training data is skewed towards problems and even though there are many stories of AI picking up false negatives the false positives have a real cost. This is longer topic, and I don’t want anyone to jump to conclusions either way. That said, for mass scale screening there’s a huge potential since the alternative is no imaging at all. Scanning for glaucoma, diabetes, or MD in eyes is a huge upside. There were some early attempts connecting imaging sensors to mobile phones, but a Korean company has done that and combined the images with an AI model. I did the scan and in 30 seconds got no false positives. Super cool. To those that know, there was not even an air puff for Glaucoma.

Glasses/Headphones/Wearables

There has been a ton of excitement this year from the work by Meta on its glasses after the surprisingly excellent Apple Vision Pro that is still searching for a scenario and practical embodiment. There’s a decided shift from AR/VR though on the floor to XR and making glasses as much like glasses as possible. I am not expert in the latest in the field, so it is super difficult for me to sift through what seem like a very large number of OEM glasses of pretty low quality and a lot of very specific components in search of buyers. That said every booth featuring glasses was crowded and in a post-covid world everyone was happy to try them on. The Luxotica/Ray-Ban booth with Meta glasses was very heavily trafficked.

This is a set of glasses specifically designed to just watch immersive Google TV:

There is a glut of new products offering basically wireless in-ear headphones with a charge case and app but touted as hearing aids. While my hearing is declining, I don’t yet use assistance, so my first party experience is limited. It is clear in the US with recent FDA changes and particularly Apple’s latest Air Pod Pro and iOS releases we are collectively on the brink of major reductions in cost and broad increases in access to hearing assistance. There were dozens of booths, of varying regulatory oversight, showing off headphones. They generally looked like AirPods in form factor. Often, they came with apps connected to transcription or video captioning tools.

There were quite a few wearable rings. It is tough to gauge the quality of the sensors and the reliance on a siloed app can be tricky. Some do integrate with Apple Health, but the Chinese origin app introduces risk for data leakage.

Smart home / Home Automation / Pets

Home automation this year was a bit of a letdown. This is a category of multiple “gatekeepers” including the legacy infrastructure of homes (switches, garages, doors), connectivity (zigbee, wifi, etc.) and the mobile platforms (that control the connectivity and gathering of devices into a “home” for permissions and access. I love this stuff, and our house is a lab. That said, this year didn’t show a lot of progress. There were fewer companies showing off the standard suite of “alarm” products (doorbell, door sensor, flood sensor, etc.). I think that is good.

Door locks made a lot of progress over the past year with some Apple Home Key locks making it to market finally (bought s Schlage essentially off eBay because you couldn’t get one at Home Depot). But Home Key is slow to deploy fully (no sharing). In some ways the locks are going backwards. The thing this year was to add face recognition to the lock. This is driven by Android primarily in China (Philips is just the distributor below). I could not get over the scale of these locks. Somehow, they turned an iPhone Face ID notch into an 18” gianormous lock. Pass.

Amazon and Kidde announced a CO + Smoke detector which is welcome for the Ring family. Unfortunately for something that most people already find ugly and want to hide, this product looks literally like a CO sensor glued on top of a smoke alarm. So, it kind of went the wrong direction.

Animals play a huge role in the Home Automation side of CES. The market for people buying stuff for pets is enormous. And it is full of love. I counted over a dozen automated cat feeders and self-cleaning liter boxes. I have a ton of friends for whom these are life savers. Our boys will not go anywhere near the enclosed litter boxes, and they refuse to eat anything that they do not see us put in the bowl (“must be fresh” they meow). They are not spoiled at all.

Birdfy is a wonderfully fun outdoor bird feeder + camera. I had one for a brief time and loved the photos, but the California squirrels attacked it en masse and tore it down from the post. They introduced several new models for more sedate areas. I love the photos. The hummingbird feeder is going on our terrace in Seattle where in the spring we get visitors many floors up in the urban setting.

The folks at Pawport spent a lot of energy designing a dog/cat door that works super well. It is an industrial strength door controlled either via a collar worn by the pet (proximity opens / closes it on a schedule) or by an app manually. The door itself is literally bullet proof and you can even install one on each side of the door for security/climate needs. It is a great execution.

Back to the kittens. We’re completely obsessed with the home gardens. These are multiplying. CES has many industrial scale gardens from Japan on display and there are an increasing number of home gardens, especially for urban customers. These aren’t just for pot anymore. They all have proprietary pods with fertilizer. We grow all sorts of stuff including kitten grass. This one is called the Plantaform Smart Indoor Garden and is powered by NASA technology! Below that is a much larger popup greenhouse sized version.

Auto / Transport / Drones

As mentioned, I am glad the “let’s just turn CES into the west coast Auto Show” has been toned down. There seemed to be little new in general across the whole of auto and transport. The booth that seemed to get all the attention was a crazy Cybertruck-looking EV with six wheels that could also contain a dual seat autonomous passenger drone also EV. None of it seemed remotely feasible in reality and was not unlike Little Nellie in some ways. IYKYK.

Here’s another one of these drone models that seems unlikely.

There’s no shortage of drones in market but the market has shifted to professional. This HoverAir X1 has an 8K camera and vision recognition so it can track you while doing adventure sports or something. It is a pretty cool form factor, and the camera quality was high. You can also pilot it manually with a phone and cradle.

Honda unveiled a prototype EV line called Honda 9. There was a passenger car and medium SUV. They seemed greatly influenced by Tesla to me. There was also a Honda/Sony prototype that featured a PS5 built into the car.

Past years saw an explosion in scooters and other micromobility solutions. The incremental innovation happening in this space is really in pedal assisted bikes. In this case the innovation is on the whole negative in my view. The bikes are getting enormous, heavy, fast, and unwieldy. Living in a hilly city I routinely see these bikes flying at car-speed down hills outside the bike lanes and essentially out of control. They have a stopping distance equivalent to a car, but the riders treat them like bicycles. They exist by threading the needle of class II and class III vehicles. Even the most legitimate vendors like Segway who follow the rules are saying that particular models are exclusively for off road use. In an urban setting where traffic and open drug use laws are not enforced, as well as any proper use of bike lanes, it is no surprise they will count on the non-enforcement of e-bike classification as well. These are most decidedly a negative.

On the other hand, much more preferred to over-powered and impossible to stop motorcycles with pedals I would love to see more potential for urban use of these neighborhood class EVs. I don’t hold out hope in the US but boy these are great.

TVs

TVs are always fun, and it is difficult to resist shopping. AI made its way into TV with lots of claimed use of AI for audio and video processing that I am pretty sure everyone reading this disables.

Google TV was there in force driving its offering, as was Roku. The primary focus was on the software side. Most of the major TVs have the Google TV app but TCL and Sony build in the software as the OS for the TV while LG, Samsung, and others have their own OS. I’ve tried both and found them all frustrating. There’s no reason to have a box requiring another plug (and potential internet connection) and wish this would just work out. Obviously, I get the complexities. At the other extreme China maker ChongHong simply offers TVs with all the different operating systems:

The primary “news” in TVs is just how big they are getting and how economical those huge ones can be. TCL is the low priced-big screen leader with 100” + mini-LED. These are beasts of course and you’re not really wall mounting them or even putting them on typical floors. But they also sell for $2000 which is kind of insane since 3 years ago the size was 100,000 custom order.

LG was showing off a transparent TV that has a screen that rises up to make it a regular TV. It is super cool, but I don’t quite get what it does. It is kind of a neat room divider. It costs like 20,000 US but is available.

For big screens the laser projector seems on the edge of uptake from pocket to 12” from wall to 160”. There was a Sony previously that the screen rose up from some sort of furniture though that model isn’t around HiSense has picked up that form factor. The inconvenience of a screen remains but the throw distance of 12” is kind of wild.

OLED is pushing down to next tier of OEM and the battle is keeping the mini led price point up for another cycle while the micro led ramps. But micro is super nice and goes to 163”. Behind the scenes is an IP battle between East Asia countries. I suspect we will start to see OLED in computer monitors as well.

There seems to be a big push to have the “art display” TVs. That is TVs with an idle screen showing photos or famous works of art. This might be an Asian market interest since in the US we tend to prefer “HDMI 1 No Signal” as the screen saver. A disproportionate amount of floor space seemed to go to these art TVs. One maker was pushing the subscription that was offered with the tv as well. Below is a whole exhibit of focused on Art TV

At the extreme low end there was a super tiny “tri fold” project that looks exactly like a tri-fold charger but is a 60” HD projector. I’ve always loved these but never know where to use one. They are amazing to see.

PC/ Gaming / Laptops

At the start of the show Dell announced a renaming/rebranding of the laptop line though they did not have a public show presence. Lenovo had their off-floor demo stations showing a laptop with a second screen that expands from the main screen. Sony was showing an accessory screen for their laptops. In general, these 11-15” OLED panels you can see (depending on feed) on Instagram ads endlessly are all over the place. They aren’t new, but the price has dropped to incredibly low levels, perhaps $100 on Amazon already.

I just don’t see more innovation in laptops happening any time soon. The industry is stuck waiting on Intel and the move to ARM doesn’t make any sense for the work involved. We’ll see what happens. I am sure we will see some new models that come with discrete Nvidia chips, but I am far more bullish on DIGITS filling the white space for innovation in laptops given it can break from the Intel dependency and deliver a unique and unmet need.

To that end there is a lot going on with “mini-PC” which are Linux/Windows devices in the Mac Mini form factor. Many tech enthusiasts have played with these. They are (also) readily available on Amazon for pretty low prices. Right now, they have terrible thermals, and all use big power bricks but perhaps that will change. Below are Mini PCs from MSI (the mobo maker) that have Intel processors. These have dual multigigabit net adapters which must meet a need I do not understand.

There are also mini-PCs running new ARM chips from Qualcomm which means they come with a promise of power. These can run Windows 11, or they can run Linux. They run Snapdragon Elite 1000 with 45 TOPs (by comparison Nvidia says DIGITS has like 4000 TOPS). Geekom showed an ARM mini as well as an AMD mini:

AMD which has been on a roll had a great partner booth with a ton of gaming machines and AI compute machines. These were kind of insane builds at the high end. At the low end there were some mini towers. These are high specified but are still pretty hot and had a good hot wind blowing out the back.

My favorite PC booth for the past few years has been Razer. They’ve done a great job building around a community and focusing on tech elite gamers. Each year they have a “project” which is an early prototype of a new product. This year’s project (Arielle) was a gamer chair that had a fanless dyson fan in the seat and back. It was pretty…cool. (sorry)

The big announce from Razer was a proprietary extension to the moonlight/sunshine remote game player protocol/app to enable remote gaming. Lots of wow factor if you want to play games on your home rig from work (assuming you can get the drivers on your PC). The work they did was to do adaptive video size though this means your home PC is offline to local console use. Here’s playing a game on a phone.

In a similar note, Intel was showing how to share trackpad/keyboard and operate across two PCs. I tried to make a video, but I got so confused as to what I was doing. Still, it is a wild PC trick though I would not recommend every downloading the drivers from anywhere and installing them unless you want malware to mainline your PCI bus over thunderbolt. It is called Thunderbolt Share and described as:

Thunderbolt™ Share unlocks ultra-fast PC-to-PC connectivity experiences that fundamentally change the way creators, gamers, consumers, and business users interact with two PCs. Its intuitive easy-to-use interface allows for more productive workflows, optimizing space and enhancing overall performance, all with two PCs equipped with industry-leading Thunderbolt technology.

The software enables users to easily and securely share screens, keyboard, mouse, and storage at the speed of Thunderbolt technology, offering exceptionally responsive screen sharing, ultra fast file transfers with simple drag-and-drop functionality, folder synchronization, and easy file migration from an old PC to a new PC.

A fun dock with apps and a clock and more:

Smartphones and Accessories

There was very little by way of smartphones. There continue to be a number of supply chain based ODMs making phones and tablets and many of these become the point of sale or delivery person devices we see in everyday use.

The accessory of the show—meaning the thing you saw everywhere and see on Instagram ads—is the tri-fold charger. It is just a travel charger that does Qi/Magsafe charging for phone, apple watch, Air Pods. One of the ODMs told me it is the best-selling charger and he’s getting requests from all his customers. Some of these are stationary made of metal and others fold up and are fabric. I don’t like these as much as the 2 in 1 that have a ring since I can easily charge and watch a movie on a flight and in the hotel, I don’t need to charge all three devices at once. It saves on weight. For the ultimate I combine this with a 65w adapter and a dual output cable and can charge MacBook as well.

I like these chargers with retractable USB-C.

The three companies doing great accessories are J5, Rolling Square, and Anker. They push the ODMs and add value in their designs. J5 is doing some great laptop docking stations. Rolling Square has interesting FindMy accessories and the best magsafe connected to laptop for video gizmo. Anker has the best stand-alone chargers. I will just caveat that last point with saying that Anker has gone crazy adding LED screens to chargers, and I don’t think we need that. One desktop multi-device charger has a whole “Tools Options” and a knob/button to pick how many watts go to which output. 🤔

Here’s a J5 dock that was super nice with a magsafe on top:

Here’s an Anker with a screen and also a super nice cable from Anker.

If you need a travel adapter (I have not needed one in 10 years but just in case) this one has done well for me and now has a 65-watt PD and keep in my go bag.

This is a Shenzhen ODM that is doing great work. I had a great conversation with the engineer who noticed I was using his battery case. He was telling me about their work at pushing FIndMy. Here are three prototypes they are selling to OEMs these days which you can soon see on Instagram ads I bet and Amazon. Here’s a passport wallet with FindMy, a TSA lock with FindMy, and a super cool FindMy built into a mini-flashlight keychain. We talked about the need to disguise FindMy devices and how using the Apple AirTags for luggage is problematic.

He also told me that the US market hates this kind of charger, and the lack of success surprised him.

This looks like the display at the Apple Store

Robots

There were a lot of robots. The big new thing this year are more humanoid form factors. Many of the ones on the floor were non-operational and several times I saw the developers anxiously trying to make them work. The robots fell into these classes:

• Arms. There were many arms for moving things around from a stationary point. The primary scenario is for assembly and the factor they all touted were the cameras and ability to quickly learn what the steps were and how to correct for misaligned parts and so on. These require a lot of work to put to use as one would expect.

• House cleaning. Still seeing a lot of Roomba clones. The big news was one that combined the mobile vacuum with an articulating arm on board to move objects out of the way. Seems like a pet fight waiting to happen.

• Warehouse assist. Particularly in Asia there were quite a few load carrying robots, carts with wheels, to move things around a warehouse. These were all autonomous and had cameras for obstacle avoidance and so on. These are all fantastic.

• Human assist. In the health area there were a few human assist robots. Two were full mobility assist for those in wheelchairs. Then there were several for legs and arms essentially for therapy or temporary use. A photo of one is below.

• Humanoid. These are like the Tesla robot. I’m pretty bullish on these and expect Tesla to do great because they are focused on building a platform and also in parallel using them in the factory. I believe that feedback loop is critical (much like Space X and Star Link, or Windows and Office). I think the ones on the floor were mostly for show and not on a product roadmap.

Gadgets

Every year I lose track of all the gadgets I see. I took over 300 photos this year. Crazy!

By far my favorite was also the biggest. It isn’t a gadget but a serious emergency response “kit”. It is MBESS – mobile battery energy storage system – and is built in the US in Baltimore. It is a trailer that contains 90kwh of batteries in a trailer that is certified for emergency response. The power can be used in place of a generator which for fire areas is really important (diesel is risky). Here’s the cool part. To be used in an emergency setting it needs to be in communication even when it no longer works. Generators carry backup batteries for this. This device has a battery backup too, but it also has the solar panel to use to power comms and telemetry even when the main battery goes down. It can be charged by power mains or by an EV charger. Conversely it can charge an EV. The capacity is about one Tesla worth. People get very confused by the panels which would take about 2 weeks to charge the battery. That’s also why that BS marketing phrase “solar generator” is so dumb. It can also be used for any remote power needs (they told me “weddings” and then I had questions). Super cool and MADE IN THE USA.

Toothbrushes seemed very big this year.

These Dyson clones were ever-present in the Shenzhen section.

Have you seen the ads on IG/Amazon for “powerful air blower” to use in your car or for your keyboard? These are bladeless USB-C powered / battery operated fans, and the airflow goes either way. There were dozens of booths with them. This one took it to a whole next level, like a Ryobi line, and added lights, blowers, heaters, brushes, and more to the same basic engine.

There was a big show presence for IoT which was heavily industrial. I wanted to end by highlighting this product from Blues, a Boston-based company making an entire IoT platform. The thing is this isn’t just any startup trying to platformize IoT (there have been many) this one was founded by Ray Ozzie (Notes, Groove, Talko, Microsoft, legend). This is really an IoT platform done right. They have developed this as a secure platform where the device talks to their service and the hardware side is a modular kit that provides varying levels of connectivity and sensors. The thing that scares me about IoT is the same as what scares me about home automation devices which is I can’t stand thinking about all the code bases sending data and all the clouds and all the exposed surface area. This is the right solution to this, and the industry should have a platform winner here. Way to go Zak (who I talked to used to work on Windows) and Ray!

If you made it this far please feel free to share this. Also love any questions, comments, or pointers on what I missed (or maybe just missed writing about). Thank you for your support.

Oh, and a final fun note, the Bill and Melinda Gates Foundation had a booth, and it was running a Surface Table.