The next 20 year tech cycle

When I first used the Web, my first thought was how useless it was. I was an engineering student at the time and had noticed this new thing called Mosaic on my school’s network. After browsing around for half an hour, I gave up and went back to work hacking into other student’s Xterm sessions.

I was already pretty familiar with the Internet. I had grown up with computers, used modems to connect to BBSes (bulletin board systems), and used various resources like Gopher for school. This new thing called the “Web” looked so lame though with its links to nowhere, wall of text, and useless sites.

I could not have been more wrong! It only took three short years before Yahoo went public and blew open the doors of the dot com revolution and the potential for the Web. At least I was not as wrong as Ken Olsen, Founder of DEC, who famously said:

Any new technology looks like a toy at the beginning. In fact, Chris Dixon, General Partner at Andreessen Horowitz, had this to day about technology:

Just think about some technologies launched over the past few decades. We saw the rise of PDA’s (personal digital assistant), 3D printers, flat screen monitors, MP3 players, and broadband networks. The first iterations of these devices like the Apple Newton, Makerbot, Microsoft Zune, and others were too wonky for the average person to use. Instead it was the geeks and hobbyists that poked and prodded around these new-fangled devices.

The same was true about the early days of personal computers. Jobs and Wozniak built clunky devices out of a garage. Gates spent countless hours on a mainframe terminal at school. This was the domain of tinkerers passionate about new technologies and their possibilities, the type that Ken Olsen did not understand. That is the thing about geeks, they don’t care what the naysayers have to say.

Which brings us to the current day with the advent of Generative AI. You cannot throw a stone in the Bay Area without hitting an “AI expert”. We have never since the pivot that we saw this year from no one talking about AI last year to having AI be the only topic anyone in tech is talking about.

If you are of the more skeptical type, you could say we have been down this road before. We have seen previous period of AI hype in the late 90’s when Deep Blue best chess grandmaster Garry Kasparov in 1997, the DARPA Grand Challenge launched the age of autonomous driving in 2004, big data emerged in 2010 allowing for massive training sets, and then a mid-2010’s spike in so-called AI startups.

Recently Sam Lessin, ex-Facebook and founder of Slow Ventures shared his views on the current state of venture capital, investment themes, and the AI craze (access the doc here with your email). Some of his points on AI are valid, especially the unhinged crowding of VC into the space and how AI will be a boon for the small business entrepreneur. We also agree there will be significant amount of value capture in this current AI cycle by the large platform players (Amazon, Google, Microsoft, etc.).

However, we have a hard time agreeing with one of his core arguments on the broad scale impact of Generative AI and the current cycle we are entering:

It is easy to miss the disruptive trend when it is first happening because it is often nerds leading the charge then hype overtaking the narrative. Generative AI first came into existence in 2018 with the Transformers paper, “Attention Is All You Need”. It was not until the launch of ChatGPT in November 2022 for Generative AI to gain widespread interest, gaining 100 million users in two months. Now the market is flooded with AI startups, massive VC funding rounds, and every tech event focused on AI.

While all of this feels exciting, what we are seeing now is not the revolution of AI. Instead, we are seeing the experimentation phase of AI. The reality is we know less about the future of this technology than any preceding leap in technology innovation, going back to the 50’s.

It helps to look back at the history of where we have been in previous technology cycles to understand where we are presently with AI and what the future holds. There have been three distinct phases of technology innovation which had massive, broad scale economic and social impact:

1. Age of enterprise computing – With the advent of the mainframe computer, businesses moved into the information age, digitizing paper-based records into centralized datastores enabling unprecedented access to data to speed and automate processes

2. Age of personal computing – The rise of the PC in the early 80’s ushered in the first consumer experience with computers, leveraging the power of compute through personal productivity software, gaming, and applications.

3. Age of cloud computing – The Web enabled even greater innovation through the power of SaaS, e-commerce, and cloud platforms which not only empowered consumers through apps, but also created an explosion of entrepreneurship across the globe.

None of these phases emerged instantaneously. Each phase experienced a lengthy period of testing and experimenting. The mainframe was slow to gain adoption with many business leaders skeptical of the value of these costly machines. The PC was purely a hobbyist toy until the Apple IIe was released and gaming consoles such as the Commodore 64 were flying off of retail shelves.

Many assume the next great leap in innovation was the Web. The true leap however came about only as the dot com hype turned into a bust. We saw this across four categories:

• Software-as-a-Service – Startups like Salesforce, Concur & NetSuite offered their software “in the cloud”, touting “No Software”

• Consumer Web – Services that made the experience for the average user stickier like YouTube, Wordpress, and Facebook came online

• E-commerce – Recovering from the dot com doldrums, Amazon led the way and startups such as Shopify, Etsy, and Groupon launched

Once Amazon made Amazon Web Services available for companies to use on-demand compute, this solidified the long-term value of the Internet for businesses and consumers.

The Age of AI computing is just beginning. We are seeing the rise of apps as “toys”, the building of the “picks & shovels” tools, and a battle for chips. New foundation models are sprouting up weekly and will continue to do so for the foreseeable future. The real value and the true new $100 billion platform players however may not emerge until the end of this decade. We are only in the start of the next 20-year tech innovation cycle.

What does this mean for you as the entrepreneur? AI is definitely where the future lies, but you are also not too late despite the hype phase we are currently experiencing. This is the time to experiment, use AI in practical ways to accelerate the growth of your startup, and to use AI as a platform for innovating. If you are planning to launch an AI-core startup, see our guidance on the AI market in a previous post.

If you are wondering how to use AI for your startup, there are tons of resources at AWS to help you along the way. Let us know what you are building, and we would be glad to help!

Mark & Basil

We get a lot of startup pitch decks sent our way. We also see plenty of live pitch events, like recently at Tech in Asia and the Orbit Startups pitch practice session.

One common slide in these pitch decks is the four-quadrant competition slide. As Peter Thiel says, competition is for losers. So we understand why founders want to present themselves far ahead of any potential competitive threats.

However, most early-stage investors review from 1,000 to 3,000 startups over the year. Based on thesis and investment focus, they will likely see a handful of startups solving the same problem in similar ways.

When you state you have no close competitors, you demonstrate to investors that you do not truly understand your market. Savvy investors will immediately identify a few startups that are at least adjacent in approach, which will put you on the defensive during investor meetings.

We recommend ditching the four quadrants, and instead segment competition in 3 buckets:

• Existing incumbents and legacy players - these are the solutions you are potentially replacing or supplementing

• Horizontal players and platforms - these are solutions that broadly solve the problem but not in an optimal way

• Niche providers - these are your closest direct competitors focused on similar markets and problem space

End with a concise statement on how you are uniquely differentiated. Your talk track is to explain how your team is best positioned to capitalize on your unique insights about the market and execute to create a competitive moat.

If you have other ideas on how best to present the competitive landscape, please do let us know and we can include in next week’s edition!

After a few whirlwind weeks, we are focusing on our virtual programming over the coming weeks. Stay posted for announcements on shows and content that we plan to release as we close out 2023.

Shout out to excellent work by the AWS Startups team in Abu Dhabi last week at the UN’s World Investment Forum! Huge impact was made with Basil on stage to share the AWS vision on Generative AI!