When new industries emerge, entrepreneurs and investors tend to ignore the hidden layers that will be key to the industry’s growth
In the 60s, all eyes were focused on the race between companies developing mainframes. In the 70s and 80s, it was the competition to win the personal computer space. Headlines in influential periodicals of the time boasted about the transformative power of up-and-coming competitors like Altos, Tandy, Wang, Zenith, Olivetti, Commodore, Amstrad, Apricot, etc, etc. Investors flocked to back these companies. Any one of them could be the next big thing.
Today, I’m willing to bet very few people have ever heard of any of these names (except perhaps Commodore, which is better remembered these days for its gaming consoles than its computers). Yet during that same era, something else was happening. Founded in 1968, the California-based Intel began producing microprocessors. Every one of those companies I listed above became a customer. And eventually so did a few other companies you might have heard of: Apple, IBM, Dell, HP, Lenovo, Microsoft… The list goes on. To this day, Intel sells a strong supermajority of personal computer and server CPUs in the world.
When graphical interfaces became popular in the 1990s, these computer companies continued competing in an ever-evolving space. But it was the 1993-founded NVIDIA that sold to almost all of those and more the components required to power their graphics (GPUs). This is NVIDIA’s stock over the past few years, three decades into the company’s life, with a market cap exceeding one trillion:
Consider all of the following companies that quietly cornered the foundational layers of entire industries and their respective supply chains, all while the world was focusing its attention elsewhere:
- Adobe for creativity tools
- Texas Instruments with electronics components
- Sysco in the food and beverage space
- Square in the mobile payment industry
- Boeing powering the expansion of aviation
- John Deere for agricultural equipment
- Github for version control
- E Ink in the e-readers industry
- Broadcom in powering components for networking products
In 2017, Anchor, the company that I co-founded, made a key strategic decision that would come to change the trajectory of the business. For the preceding three years we had tried with limited success to “democratize” the podcasting industry by focusing on the end-consumer use case. But despite our ambitions, the numbers told a different story.
The key decision we then made was to stop focusing on the listener proposition entirely and move away from consumer products toward the infrastructure layer. This was the hidden layer of podcasting, the one that, while key to its growth, was rarely the topic of discussion amongst founders and investors at the time.
We aspired to neither create content nor provide it directly to listeners. But between those two ends of the industry were the forgotten middles ripe for innovation and rapid growth: podcasting creation, hosting, and monetization tools. Today, products serving those needs are numerous, with many companies competing for a share of the pie. But six years ago, when the podcast industry was just on the cusp of experiencing rapid global growth, that hidden layer was an area very few entrepreneurs, VCs, and media were talking about. This, we believed, was where the gold was hidden (in plain sight too).
The Hidden Layer of an industry is that in which goods or services exist to power downstream consumer-facing goods and services. You can think of these as essential middle-of-the-supply-chain companies, although the term “supply chain” often has implications for the conversion of raw materials to finished goods. You can also think of it as infrastructural or perhaps B2B products, although examples like Adobe or Boeing or even Anchor showcase that the terms infrastructure and B2B aren’t always appropriate descriptors.
Several years ago, as the crypto space was booming and all of the focus was on which one coin might rule them all, it was Coinbase that took the hidden layer approach. Agnostic to which cryptocurrency might prevail or which use case might become common, they focused instead on connecting the two ends of the industry. In the space industry, it’s the launch service providers that are betting not on who will control the sky but on the fact that someone will, and their aim is to work with all of them. In renewable energy, it’s the companies integrating with the grid. In AI, it’s the companies betting on data processing and computing capabilities. In the EV field, it’s battery manufacturers.
If you’re a founder building consumer-facing application layer products, competing in a hot (and hotly competitive) space, consider the opportunities that might lie right behind the curtain, in the shadows, where a first mover advantage to serve the industry (rather than end-consumers) may one day prove tremendously valuable.
At its core, there are three reasons I see why hidden layer strategies are naturally advantaged when an industry first emerges. And these exist in addition to the fact that, due to the nature of virality and attention, the farther you move away from end-consumers, the less competitive the space becomes.
If you’re thinking of starting a new company in a space that is not yet clearly defined, where there is a sudden, fierce race for consumer attention — or if you’re investing in one of these fields — consider the following three hidden layer advantages you might be overlooking:
- The Rising Tide
- Established Connections
Those building applications, not infrastructure, are competing for a slice of an as-of-yet-undefined pie. Remember that we’re looking at industries that are up-and-coming, the ones that have not yet found maturity or product market fit, the ones early enough in their adoption curves that the competitive consumer proposition is guaranteed to evolve. We’re still at that stage today with AI, VR/AR, tele-health, 3D-printing, blockchain tech, etc. We probably are just moving past it in the electric vehicle race.
Yet despite their ambiguous future states, one thing is clear with all of the industries I mentioned above: they will be substantially larger ten years from now than they are today. Many of the companies doubling down on specific consumer propositions won’t be around in a decade (recall all those PC companies I mentioned up top). But whoever is around will likely be serviced by components or content or infrastructure provided by a player operating in a hidden layer.
Hidden layer companies benefit from the rising tide and, in particular, from their relationships with companies farther downstream. Intel did not know in the 60s and 70s whether Altos, Tandy, Apricot, Apple, or any of the others would eventually beat out the rest. It didn’t need to. It simply needed to have conviction around the growth of PCs and serve as much of the downstream market as it could to ensure it could ride the growth of the winners.
Once Anchor established itself as a hidden layer company in the podcasting space, it was substantially easier for us to expand to other parts of the ecosystem. For instance, we could become the supplier of advertisements that allowed creators to monetize. We could invest in producing original content. We were even able to scale peripherally to support other types of content creation, like video.
From the hidden layer, it’s significantly easier to expand your business to other parts of the industry. You already have the relationships, the industry expertise, and the visibility into what’s worth the investment.
Intel, for instance, became the Intel we all know by building microprocessors for PCs. But today, a third of their business powers server and networking hardware used in data centers around the world. Expanding to cloud infrastructure was, in many ways, a de-risked investment, with clear upside and a healthy sibling revenue line to help sustain it during the early days.
Adobe is scaling into the generative AI space with incredible products like Firefly, but they’re doing so from within an existing user base and a network of creative customers that it built off an entirely different line of business. Keep in mind that the end-consumer of creative content is not the creator using Adobe products. It’s the downstream customer of that creator’s work. While it’s an application layer product developer, Adobe products live in the hidden layer of many creative industries (photography, music, movies, design…)
One thing is certain about emerging industries: In several years’ time, they’re bound to look drastically different. They always do, and no predictive model will accurately reflect those futures. Yet despite the evolving nature of the consumer propositions or the emergence of new technologies that will power innovation, it is highly probable that the majority of the players in the space will stay the same. I can’t tell you which companies will succeed and which will fail, but I know that most of those who will succeed will be the ones who joined the race early and were able to evolve with the industry.
And regardless of how much these companies change their products, inertia guarantees that the majority of their vendor relationships will stay the same too. In other words, the connections established early on remain strong as industries evolve or players expand.
And who in an industry has the most established connections to a broad range of consumer-facing companies? Those in the hidden layer who began serving those same companies early on.
Consider Intel once more. Cloud computing wasn’t fathomable when Intel entered the mainframe and PC markets many decades ago. Yet when large tech companies did transition their businesses toward the cloud, a familiar partner with whom they had worked for decades was soon available to power that infrastructure.
The term “hidden layer” has tremendous significance in the AI space. In neural networks, hidden layers represent everything between an input (your prompt to ChatGPT or DALL-E for instance) and the output (the text or image response). They’re where the actual work happens, the foundation without which the apparent, downstream, consumer-facing magic can’t happen. (It’s also where, if the output makes no sense, the mistake was likely made; I explored this topic in last week’s Z-Axis article.)
There are many fascinating analogies to draw between how neural networks work and many other fields, from business to politics to decision making and mental health. But those are articles for another day…