What will the 2020s hold?

Looking back on the 2010s as a busy adult, it is hard to see much change.

I started the decade with my first iPhone, an iPhone 3GS. I don't recall if I got that iPhone in 2009, or 2010, but it was somewhere around there. And I was a smart phone user since 2005, when I bought a Palm Treo, with a data plan, and enabled my corporate email (via IMAP) on the device. I even bought software to allow me to tether the phone (via its data sync cable) to my laptop. Prior to that, in 2003, I was using SMS text messages on my GSM phone to communicate internationally. At the end of the decade (literally the very end), I had an iPhone XR.

My work tools did not really change. I started the 2010s with a Macbook Pro, and ended with a newer Macbook Pro.

Microsoft Windows went from Windows 7 to Windows 10. My home PC did not change much, an old flat screen monitor died, so I bought a new one.

My primary home TV did not change. It is still the same 1080p Sony SXRD rear-projection I bought on Black Friday in 2006. My basement TV went from a 30-inch 1080i CRT to a large 1080p LCD (bought second-hand).

While I went through four iPhones in the 2010s, I have the same 80GB iPod Classic I won in a work contest in 2006, which I use for my car audio.

I bought a new car in early 2010 (a 2009 model year car), and drove it for most of the decade, replacing it in early 2018 with a used 2015 model year car. The old iPod integrates just fine with the newer vehicle.

What did change? The single biggest consumer change I saw was the end of traditional media. Let me unpack this.

I started the decade still a member of the Air Force Reserve. As a flier in the Air Force, I had to carry considerable paper manuals. Most of these were published in PDF format, but we had to carry hard paper copies. I was curious about e-reader technology, and tested Air Force PDFs on the Sony Reader, which supported PDFs. Unfortunately, the resolution of E Ink e-readers was not good enough to render an 8-1/2 x 11 inch PDF page filled with small font text. However, in 2011 I purchased an iPad, and I found the iPad did render PDFs adequately. A couple of years later, the Air Force standardized on the iPad 2 as its Electronic Flight Bag. A single iPad replaced several large binders. While e-readers have had mixed success when compared to a handful of personal books, clearly they have an advantage for reference libraries.

The second example is when we purchased a minivan in late 2014 in anticipation of the birth of our second child. My manager at work told me: "Don't get the DVD player option. It is not worth the additional $7,000. Instead, go on Amazon on Black Friday and buy two cheap Android tablets. And sign up for their free kid app of the week." So for less than $100 for two tablets, and another $100 or so for headphones, cases, and car mounts, and another $50 for DVD ripping software, each child can watch what they want to watch. And now play games as well.

Obviously, the 2010s were the decade of Netflix, where Netflix's original DVD based business faded into irrelevance, and its nascent streaming service became an industry changing phenomenon.

Audio went through more changes. From the idea of digital downloads replacing traditional CD sales, to the emergence of streaming and the end of stand-alone MP3 players. Certainly, the smart phone played a role in this transition, as did smart speakers.

We went from renting DVDs to video streaming, and from buying CDs to downloading digital music to dedicated MP3 players which docked to speakers for the home, to audio streaming to smart phones and smart speakers. Media is all but dead.

The second most significant consumer phenomenon of the 2010s were "ridesharing" services Uber and Lyft. These services simply were not possible prior to the rise of the smartphone, and were the most disruptive business of the decade.

The third major consumer trend was the decline of retail, a.k.a., the "retail apocalypse". Everywhere during the 2010s one could find abandoned storefronts. This was not a new trend, as the decline of malls started in the 1990s, but a second wave started in the aftermath of the 2008 financial crisis. Specialty stores were obsoleted by online retailers, causing brands such as Toys "R" Us, Borders Books, and Payless ShoeSource to close, but also department stores like Sears and J.C. Penney to shrink their presence considerably. This trend is so significant, most cities are developing plans to deal with declining retail centers.

In the business area, the mainstream adoption of cloud computing was the biggest trend, followed by the mainstream embracing of mobile technology.  Cloud computing officially started just prior to the 2010s decade, as Amazon Web Services EC2 came out of beta in late 2008. By mid-decade, mainstream adoption had started, and by late in the decade, most organizations were adopting cloud computing. In that time, the concepts of IaaS and PaaS had been refined into industries of their own. Also, software as a service (SaaS) was mainstreamed during the 2010s, with Salesforce and Microsoft's Office 365 driving a large portion of that segment. The decade started with significant on-premises infrastructure refreshes, driven by Intel's significantly improved Nehalem processor, the rationalization to x86 as the defacto instruction set architecture,  the emergence of Linux as a viable alternative to UNIX, Windows Server 2008 as robust enterprise grade version of the Windows NT line, and the widespread use of 10Gb Ethernet networking. The decade ended with many large organizations getting completely out of the data center business, keeping what was required to be capitalized, owned, and self-managed in colocation facilities, and pushing everything else to opex SaaS, IaaS, and PaaS platforms.

The second biggest trend was the embracing of mobile. As Apple with its iPhone and the Google Android platform became a defacto mobile duopoly, and as smartphone penetration increased, a similar push to the late 1990s and early 2000s for businesses to get on the web emerged in the 2010s with mobile. Banking apps replaced ATMs, and airline apps replaced kiosks. Even ordering coffee at Starbucks was accelerated via mobile apps. While mobile apps are gimicky for some businesses, they are useful for many others, and critical for a few. And while most use cases for mobile were customer self-service, some businesses leveraged mobile for their own employees. Chick-fil-A's use of tablets to take orders and payments in the drive through during peak hours, airline flight attendants using mobile devices to process payments in flight, and of course products like Square, who allowed any independent business with a mobile phone to process credit card payments, and services like Uber and Lyft which are built entirely around the mobile phone ecosystem of location services and payment processing.

So what will be the trends for the next ten years? I really can't predict the consumer side of things. After all, I have a 13 year old rear projection TV, and just replaced a 2014 model year iPhone with a 2018 model. I bought a 2015 model year car about two years ago and intend to drive it as long as I can. I was a later adopter of Netflix streaming, finally joined Amazon Prime in 2019, but did subscribe to Disney+ within weeks of its launch.

My first prediction would be streaming surpasses broadcast and cable/satellite. Some channels will likely die off and fade away. Every media company will offer a streaming platform. People will expect an "all you can eat" model from the streaming providers, and there will be a war among the providers as to who can provide the most content. I also expect more ad supported "freemium" models to emerge. Related to this, I expect to see more audio streaming. Ad supported streaming has a strong potential because internet advertising offers the highest level of targeting and customization, which is something broadcast is not capable of providing. Streaming also offers the potential for interactivity, so the idea of interactive television could finally become a reality, with the viewer able to choose from different content rated version, and select from multiple story arcs. The other likelihood is, with all of the new streaming services, and the competition between them, Netflix's market cap will come down to Earth in the decade of the 2020s.

My second prediction is some autonomous vehicles will become mainstream by the end of the decade. What form I do not know. It may be limited to some kinds of minibuses on fixed routes in dedicated lanes, or it may be Level 4 or 5 autonomous automobiles. If Level 4 or 5 autonomous passenger vehicles become mainstream by the middle of the decade, Uber and Lyft will survive and likely thrive. If autonomous passenger vehicles do not become mainstream by the end of the decade, Uber and Lyft will disappear in current form. What is important is even if automated vehicles are limited to certain lanes, the idea of an autonomous minivan carpool-like service would be disruptive compared to the alternative of less frequent city buses. I think the significant thing to consider is, automated vehicles are likely to first appear in a form we are not expecting. Perhaps automated Bus Rapid Transit in places without existing mass transit (or mass transit driver unions).

In the business space I occupy, IT, I have better ideas, simply because I get exposed to my employer's roadmaps and hear about the strategic IT plans of customers.

My first predicted trend is edge computing. The push to the cloud will over-reach, and the overreach will be first seen where low latency and high bandwidth are needed. This will drive edge computing, if for no other purpose than to process and compress local data sets before uploading them to the cloud.

The best analogy for edge computing are content delivery networks (CDNs) today. If one looks at the history of CDNs, they originally were created to deal with the limited bandwidth of the internet, and limited computing power of internet servers. The appeal of CDNs in the era of the early dial-up internet faded as broadband penetration increased, but they would become critical again with the rise of downlodable and streaming media.

I mentioned Netflix and the rise of streaming media replacing traditional DVDs and CDs. What makes Netflix unique is its content delivery network, Open Connect. Some people love to point out Netflix was an early and aggressive adopter of Amazon Web Services. But when you watch a show on Netflix, it is not streaming from AWS, or from AWS' content delivery network, but from Netflix's own CDN. Netflix negotiated with ISPs to place their CDN appliances in the ISP's facilities. In most cases, this means the local telephone exchange facility for DSL based services, and the cable head-end for cable modem based services. This means Netflix is only streaming over the proverbial "last mile" of the Internet. Programs are pushed out to the Netflix Open Connect appliances ahead of availability. This is why when a new season of a show drops, millions of people can binge watch it with no slowdown of the Internet.

At the end of the 2010 decade, Disney launched Disney+. Prior to launching Disney+, Disney had to contract with existing CDN providers to ensure they could support the expected customer base. While the specifics are confidential, it is speculated Disney contracted with many CDN providers to ensure they had content as close as possible to their customers.

Video streaming providers and content delivery networks provide an important archetype for the future of edge computing. In additional to content delivery networks, there will need to be compute delivery networks.

Given today's Internet bandwidth, CDNs provide more value as a latency reducer. The latency for a web page, or simple streaming videos can easily be measured in the tens of milliseconds. However, edge computing will likely require sub 10 millisecond latency, and perhaps sub 5 millisecond latency. It will not be possible to simply locate compute resources in the same facility as existing CDNs.

It might be an ideal use case for existing local data centers, such as colocation facilities and existing corporate data centers. Alternatively,  edge compute networks could be built in conjunction with 5G network facilities. 5G and edge compute facilities might be to the 2020s what "cloud plus mobile" was for the 2010s. 5G's limited broadcast distance will require many facilities. These facilities will connect into backhaul fiber networks. These new fiber networks will be installed throughout urban and suburban areas. It would be easy to connect an edge compute network facility to the 5G backhaul network. We have ten years to build this out a combined 5G and edge computing infrastructure, so I do expect it to look new and different, not bolted on.

My second prediction is AI goes mainstream. There was a lot of noise about AI in the last decade, peaking in 2018. Hype was bracketed with warnings of another "AI Winter". NVIDIA's market cap ran up to an incredible high, only to lose half its value. Companies bid up to seven figures for newly minted data scientist PhDs. But not much was visibly produced by AI. Most if it was invisible, things like fraud detection. That which was visible was not promising: people killed by self-driving automobiles, the Cambridge Analytica scandal, disturbing deepfakes, etc.

Part of the problem is we are still in the early days of AI. However, Moore's Law continues, academic efforts around machine learning and deep learning have expanded, software continues to improve, and companies are becoming more experienced in AI, what it is good for, and what it is not currently good for. This will lead to a rationalization, a focusing, which will produce usable results. Many of these uses cases will be simpler than the hype of the last decade.

The other thing which is likely to happen, which has not happened yet, is the commercialization of AI services. Today, each potential user of AI has to build their own AI environment. Sure, they can leverage cloud providers for the GPU infrastructure, and they leverage the open source tools, but they have to employ data scientists to build the training models, prepare the data, train the models, test the inference, etc. The idea of a third-party AI organization specializing in particular use cases, such as  retail loss prevention, coming in and doing a complete AI based solution from data identification to data optimization to training to implementation of a final product is yet to be mainstream. Once it is, we will see more of deployment of AI. I expect data scientists employed by large retailers and large banks to create startups focused on implementing AI solutions.

The last IT related prediction I will make for the next decade is the idea of "everything as a service". The public cloud providers have changed the way enterprise IT customers consume resources and services. Many customers never implemented significant charge-back or show-back mechanisms to measure the cost of providing services. The idea of pay for use on premises IT was a feature of the mainframe and has been explored for decades in the open systems space. AWS's on-premise Outposts platform can be consumed either in an opex or capex model. I have seen some customers push back on opex models (software subscriptions) in favor of capex, and I have to wonder if it may be in part due to public cloud and SaaS opex expenditures consuming their opex budget, leaving IT with only capex remaining. If this is true, I expect budgets will evolve over time to a more opex heavy model. If some customers prefer opex models, and some capex models, this means vendors must provide both. I think we will see this with both software and hardware.

Regardless, I expect we will see more change in the next ten years than we saw in the last ten years. The pace of change of technology is accelerating. It is hard to notice because we are conditioned to the current time, and we are conditioned to change, so we only see change in retrospective.