Shutterstock

Predictions for Tech in 2023

One year ago I tried to predict some tech trends I thought would take off or I wanted to take off in 2022. Let’s see how close I got. I’d also like to double down with some predictions for tech in 2023, seeing as a lot has actually happened in 2022. It has been a busy year for me in addition to writing a book so I haven’t had nearly as much time for Medium lately but I’d really like to see how I did predicting 2022.

Six Predictions for Tech in 2022

Recap 2022

First let’s see what I bet on for 2022 and how I managed. Some of these were premature. Some far exceeded what even I thought. Also there were other topics I didn’t cover such as the surge in GPT-3 and AI for major advances in chat and image generation. There was also an extended shortage of chip availability due to logistics delays and war in Ukraine. Let’s have a quick breakdown:

1. MISS: Quantum Photonics will break its first cryptocurrency. 😕
   This is the most disappointing prediction to myself. I was hoping 2022 would at least see the introduction of quantum co-processor cards similar to commodity GPUs. I still have hopes this will come from companies like Xanadu and Quix but it seems they are a ways off and still exist only consumable as a black box via cloud services. I would like to punt this prediction for next year but we still haven’t seen public instances of quantum breaking SHA256 or cracking a minor or insecure cryptocurrency hash. The largest quantum computers have increased in size with IBM reaching 433 Qbits. Hopefully this will double by next year but most likely this kind of device will not be portable for quite a while. I call this one a miss for 2022.
   [Update 5-JAN-2022: Just 5 days late researchers in China announced a possible Quantum break of RSA.]
2. WIN: Image recognition will enter the third dimension. 😀
   I was ready to concede on this one but the brilliant folks of OpenAI have just managed this in the last week of the year. OpenAI has demonstrated early versions of this via an app they’re calling Point-E which does exactly this. It’s in early stages and able to generate convincing approximate point clouds from two dimensional images but it’s just a matter of time before old photos and films are remastered using 3D model extrapolation. One thing that took me by surprise was DALL-E, the OpenAI text to image tool which produces amazing renderings from text descriptions and has become controversial in the art world. I call this a partial win at the very last minute and think it will improve in 2023 by leaps and bounds.
3. MISS: Year of the Linux Desktop. 😕
   Wishful thinking has always tipped my bias on the Linux desktop. As a former Red Hat I continue to use and support Fedora as my daily driver. It keeps me ahead of the curve in enterprise Linux releases and also active in testing the latest upstream projects. Unfortunately it has been another year of the community completely replacing long-lived projects that have finally become stable. Over the years we’ve seen stability sacrificed to rewrites of some major projects which have subsequently stunted desktop stability and hindered adoption from most users. Past replacements include systemd for init, PulseAudio for sound, wayland for Xorg, and now Pipewire for PulseAudio. All of these replacements come with their growing pains and Pipewire audio has cleverly localized audio to a user-based service instead of a system-level service only configurable by root. Unfortunately it completely breaks quite a bit of Bluetooth and HDMI audio sources in my experience, and even caused early Zoom and other software releases to error claiming that no PulseAudio service was detected. Luckily PulseAudio is still a rollback option thanks to the helpful packaging teams at Fedora. There is some great news though. The clever folks at Valve have continued to develop Proton support for Steam. This Windows compatibility layer on top of Wine is mostly to support their Steam Deck portable devices to play Windows games on the Linux-powered Steam Deck. This may feel like a bit of a false positive since it’s really enabling Linux to run Windows software. This is particularly interesting as Microsoft is all but killing off the Windows desktop via it’s unpopular Windows 11 release, which introduced unreasonable hardware requirements and forced ads onto users that were perfectly happy not upgrading at all. Chrome OS has also increased its market share which is good for these cost-effective devices. Have a look at current Desktop market share as new data becomes available. For now I’d like to call this result underwhelming for 2022.
4. WIN: RISC-V will explode. 😀
   This is the most surprising of my predictions. There isn’t a whole lot of high profile news about RISC-V, the open source alternative RISC architecture project. The truth is it is advancing more than I imagined. I remember cross-building and testing early releases of HashiCorp projects for RISC-V using QEMU emulation. The compatibility was really impressive and our projects seemed to run without issue. Finally this year we’ve seen the introduction of some development boards by SiFive, similar to the Raspberry Pi form factors. I managed to snag one of the early releases after paying quite a bit after import duty. It functions pretty nicely for development and subsequent boards have been released with more power at a reduced price. The much bigger news is how RISC-V is permeating public cloud. With hyperscalers making their own ARM hardware for a while, ARM faces some healthy competition from this free and open alternative. If hyperscalers can get margin cost of production down to less than that of ARM hardware, this may go even further. The big thing to watch for now is AliCloud, which is preparing to roll out its Wujian 600 cloud offering powered by RISC-V. As tech sanctions have hit the east hard, RISC-V’s open source nature becomes the obvious answer in a market that needs to produce industry-standard chips for devices and cloud. I’ve been on the waiting list with AliCloud for almost a year now and I look forward to test driving it as soon as they let me. The good news is I’ve still tested all of our bits on RISC-V via emulation and now physical hardware and I’ve seen no issues as of yet. Performance per watt and cost effective hardware looks promising if mass production can increase. I call this a surprising win and expect to see it surge even more in 2023.
5. MIXED: Hybrid Cloud will surge. 😐
   This too may be a bit of wishful thinking. Hardware is still relatively cheap even with inflation climbing like crazy. It seems there is a growing adoption of purely container-based orchestration for private cloud. Many Kubernetes distributions like OpenShift are being sold as private cloud platforms but in my own mind containers are still only part of the solution lacking VMs and traditional workloads. Rather than make container orchestration a feature of the broader solution, vendors are adding virtualization as a feature within container orchestration. This seems unnecessary to me as there are already plenty of virtualization offerings that are fully mature. We see Red Hat sunsetting one of their most underrated projects, OVirt/Red Hat Virtualization in favour of moving libvirt orchestration into OpenShift. This unnecessary move is designed to drive OpenShift adoption but the truth is that Red Hat Virtualization had finally became a stable and viable competitor in the hypervisor space. Meanwhile VMWare dominance is not going away and libVirt control via container orchestrators is a misplaced use case, missing plenty of libVirt’s most useful features such as live migration and hot add/remove RAM/CPU. Red Hat will continue to succeed with OpenShift and other Kubernetes distributions will gain ground as more adopt but to sell a container orchestration as a private cloud solution is misdirection. We see continued service mesh solutions based on Istio and Envoy proxy which only give connectivity between nodes within a single cluster, missing out on the hybrid or multicloud use case. HashiCorp’s own Consul continues to be a great solution to hybrid and multicloud service mesh along with multicluster connectivity where federated clusters are not supported. For myself, the ultimate hybrid cloud solution treats traditional workloads, virtualization, and container orchestration at the same level. Overall I would say this was a valid prediction for 2022, but not in the manors I predicted.
6. MIXED: Battery technology will double efficiency. 😐
   Another wishful thinker for 2022 was the efficiency of battery technology. Many innovations have taken place in 2022 but not many have hit the markets just yet. Plenty of Lithium alternatives are currently in testing and patent processes. Everything from Sodium to bricks to water have been used as batteries in 2022. The scarcity of Lithium will be its biggest limiter but Lithium is still king. After four years at HashiCorp I was happy to replace the battery in my Dell XPS laptop rather than go with an expensive full hardware refresh. Just $40 later I had a good as new laptop with fresh hours worth of battery life. Other than the dreadful nosecam and early-stage coil whine on this machine, Dell has done a phenomenal job with the XPS series and I’m really impressed with my now 5 years of reliability. In the end I will call this prediction a partial win for battery efficiency, but none of it has hit the market quite yet the way I had hoped.

What about 2023?

Now let me try to guess what happens in the year ahead. These opinions are personally my own. What do I expect or look forward to in tech for 2023?

Even experts can get things wrong, so how about amateurs like me?

1. The Metaverse will implode, taking Zuckerberg with it.
   The Metaverse continues to swallow massive amounts of money for a concept that has failed time and time again in the past. Proponents will insist that today’s superior technology will succeed where earlier attempts have failed but this is the same excuse used by previous generations. Tech is relevance is always disappointing in hindsight, and the fact they started with 1990s graphics in their rollout was a double dose of fail. This continuous drive to push a product that nobody asked for will result in major change of Meta leadership, which refuses to reverse course and conceded defeat after already squandering so much capital on the project. Every attempt from VirtualBoy to Google Glass any consumer head-mounted augmented reality introduced has been abandoned for no commercial viability. You can’t fix a product with no demand by throwing endless R&D budget at it.
2. Crypto exchanges will continue to collapse.
   The concept of copy/paste soft currencies were one dangerous trend. A whole new danger emerged in 2022 when it surfaced that millennials with no memory of the Glass-Steagall Act in T-shirts were running hedge funds with customer deposits. The interdependence between these exchanges and funds presents a house of cards, and I doubt this will be the last shocker within the community. All the freedoms from regulation presented by decentralized finance also present the downsides of no bailouts, FDIC, or safety controls. Those who lose their life savings have no recompense and stakeholders will fight for the leftovers. I have zero financial position in this space and I don’t intend to.
3. Semiconductor shortages ease🤞.
   Hopefully with the crypto GPU craze crashing down and logistics problems easing I am hopeful a re-balancing of demand will leave foundries able to prioritize demand according to what people actually need rather than top-paying crypto mines and bulk buyers. Even the humble Raspberry Pie has been hard to find, available only at scalping resellers like Amazon, which have bought up inventory and relisted the devices at 2–3x original prices. It’s time to put consumer prices back in line. I suggest people boycott second market resellers and markup scammers of all things GPU and SBC which are suddenly 3x expected cost. GPU shortages were supposed to ease in 2022. It sure didn’t feel like it. Hopefully 2023 will finish the job.
4. Project Caviar and AV1 will complete the free media vision.
   Over the past few years there has been a quiet revolution brewing in open source media. Image, video, audio, and data compression software that has been around for decades since the days of analog video and sound are being usurped by free and open source software designed for today’s world of 4K streaming and large payload internet transmissions. The Alliance for Open Media is a nonprofit group with members from the biggest names in tech. Patent trolls and litigation have made using proprietary licenses a complex mine field. Also relevant is Xiph, a similar foundation for the open source audio codecs Ogg and libOpus from Xiph, the open source lossy audio codecs of the future. Google, Apple, Netflix, Microsoft, Amazon, and others are streaming media around the world and are subject to longstanding patent fees every time they use audio standards from Dolby, image/video standards from the MPEG group, and more. Most end users probably won’t notice these technologies but they are definitely beneficial to media consumers in the long run. As digital media goes higher resolution and higher frame rates bandwidth becomes critical for both consumers and networks. When JPEG enabled lossy compression of digital images it revolutionized digital photography. That was 30 years ago. Google has since advanced the entire science for free with WebP, which combines the features of GIF, animated GIF, and JPEG all in one standard which saves about 50% more than JPEG standards for bandwidth savings. When early lossy MP3 audio compression first took 700MB stereo audio CDs and compressed them to about 10% of their size it was revolutionary. Newer audio compression can do much more than that, including 5.1 and full 3D surround sound for home theater in even smaller storage requirements. When video compression codecs like x264 took 5GB DVDs and compressed them into 10% of their size it was revolutionary also, and these became the standard for digital cable and satellite networks. Hardware decoders and encoders made this option ubiquitous but x264 and successors have largely been based on 20 year old algorithms and parameters. Early open source alternatives by the AOM include VP8, VP9, and now VP10 which was released AV1. These codecs are designed from the ground up using modern version of classic algorithms and the trusty DCT, larger resolutions in mind, and algorithms tuned for today’s hardware instead of 1990s limitations. Using WebP for images, AV1 for video, and libOpus for audio can compress entire box sets of HD media with about 25% of the equivalent storage of JPEG, MPEG4/x264, and MP3/AAC. The reason these foundations put this effort into free software isn’t entirely selfless. They stand to save massive amounts on streaming bandwidth and patent licensing while doing a great service to the community and even providing this valuable IP to potential competitors. Facebook has also gone so far as to intoduce Zstd for general lossless compression as an extremely powerful alternative to deflate implementations like Zip, Gzip, Xz, RAR, and more. Thank you to the foundations. For my own taste, JPEG, GIF, x264, MP3/AAC, zip/gzip, and other 30 year old compression tools are already deprecated and not worth using for today’s compression. The next step requires hardware vendors to add support. Hardware decoding of AV1 has been available in GPUs for a while now. Encoders are just now arriving. The next step requires full support in mobile devices, which is currently lagging and requires software decoding. Every year I archive my phone camera data for backups. Compressing the images with WebP reduces storage requirements to about 60% of original size and compressing video with AV1 now results in videos about 20% the size of their original with perfectly reasonable quality. What’s not to like?
5. Tech infrastructure budgets will be squeezed.
   A lot of large orgs are already cutting back on infrastructure and engineering staff. OPEX budgets will be under the microscope. Many cloud migrations are the result of perceived CAPEX waste. Anybody with any cloud footprint probably struggles with cloud waste. The key to cloud waste is accountability, which requires an actionable audit trail. Infrastructure as code deployed with a gitops workflow is arguably the best strategy for this. To stay relevant in a tight market, make sure to keep your skills up to date and demonstrate value within your organization, including waste reduction and efficient use of resources. The most important skills involve prudent security. Sometimes the right answers are not what a customer wants to hear. Whenever a customer asks how can they be certain that nobody can compromise their public cloud infrastructure my answer is simple. The only guaranteed way is to not have publ
6. OpenAI will become a political battleground.
   OpenAI has come further faster than I ever imagined. I figured it would take quantum computing advances to get as far as we are today with AI. The fact that GPT-3, one of the best frameworks was being developed with open source licensing was an even bigger surprise. Financially backed by visionaries including Sam Altman and Elon Musk, the future for this group looked bright. It looked bright to Microsoft too, who invested $Billions and then established an exclusive license for it, blocking access to a project with “open” in its core principals. Elon Musk was quick to point this out. Usually a collaboration like this will turn itself into a foundation before people fight over control but it sounds like Microsoft beat the community to it. The future of this tech is very bright so long as it doesn’t become a fight for control. I hope I’m wrong but I think there will be objections and controversy surrounding this topic in 2023 and I hope they don’t interfere with progress.

We’ll see how this year plays out but these are my picks so far and it’s obvious they aren’t all optimistic. I hope for the best outcome in all cases and to be proven wrong in every instance of pessimism. Below I’ve included some of my favourite and most relevant Medium links for the topics above. Medium doesn’t seem to allow embedding links or images easily within lists. Colt McAnlis does a great series on compression for Google including their media contributions which I recommend watching.

Colt McAnlis - Medium

How WebP works (lossly mode)

Netflix Now Streaming AV1 on Android

Why Facebook Will Fail at the Metaverse

The next Crypto collapse, who’s next?

What does labor shortage in semiconductor means to the world?

20 Entertaining Uses of ChatGPT You Never Knew Were Possible

Using ChatGPT and terraform