Decentralization Is The Future Of Software

    In tech investing, Unicorns are defined as private companies with a valuation of more than $1 billion. According to PitchBook, as of November 2021, there are about 925 such businesses in the world. The cumulative value of every unicorn company amounts to a total of about $3 trillion USD. This number is inclusive of enterprises in the US, Europe, China and elsewhere across many sectors, such as AI, internet services, data analytics, fintech, supply chain and more. As such, the list reflects a truly global, cross-sector view of every promising technology company that has achieved some modicum of scale.

    In contrast, the five largest public tech companies are Google (Alphabet), Tesla, Apple, Amazon and Microsoft. As of December 2021, their combined market cap was about $10 trillion. In other words, for 30% dilution, the five largest tech companies could buy every tech unicorn on the planet – nearly a thousand of them – in their entirety.

    This may be surprising to some, but it should not be. In all sectors of the economy, the fundamental law of gravity applies almost as universally as it does in the domain of physics. Power and resources are magnetically attracted to places where they are already most abundantly present.

    What I seek to explore in this article is why this phenomenon occurs in the technology sector. How do five large technology companies accumulate three times the mass of a thousand of the best technology companies across all sectors and all geographies? What are the underlying mechanisms that drive this level of concentration and dominance? Is it about technical merit and product excellence? Or are there other factors at play?

    An additional exploration is whether such concentration matters. Is it a good or a bad thing? And what does it portend for the future of innovation in the US, and globally?

    To make progress on the first question, it’s important to first analyze the very broad range of business areas in which these tech companies are engaged. For example, which areas present the highest growth for them, and how do they capture the greatest share of value from the market? Microsoft, Amazon and Google derive huge growth from their cloud businesses. For these tech giants, cloud revenues are increasing 35-45%; an incredibly high rate of growth keeping in mind the already significant base numbers. For example, Microsoft's Azure cloud revenues for Q1, 2021 were $17 billion.

    The four largest tech companies are excellent operators of large infrastructure due to their scale. They leverage this scale to reduce the cost of every increment of data center capacity they add. They then sell this capacity to other companies. Servers, power, cooling, data center real-estate, bandwidth, the software infrastructure and DevOps systems necessary to keep the “cloud” running represents significant capital expense, but - like most things - are cheaper in volume. In many ways, this shift to the cloud throws a wrinkle in the conventional assumption that in the software industry it takes very little investment to get a company going, and that inexpensive distribution allows great reach. It is true that you can indeed develop software products with a small team and utilize open source infrastructure to speed up the process. But when it comes to deploying this software, things are hardly inexpensive. Most startups end up choosing one of the top three cloud providers as the platform on which their software runs. Since these tech giants have the volume, they also command pricing advantages and have the capital to develop proprietary hardware such as application specific integrated circuits (ASICs) for cloud-based machine learning. In this domain, advantage isn't just about innovative software, it is about massive amounts of concentrated capital. And with each round of new startups that bring their exciting new products to the big tech cloud, these giants grow ever larger.

    Bob Metcalfe, Ethernet inventor, Xerox PARC researcher and research fellow at SparkCognition, famously said that the value of a network increases roughly exponentially with the number of nodes in the network. This observation is now known as “Metcalfe’s Law”. As we’ve seen, large cloud providers benefit from the network effects of the applications running atop their infrastructure. But is this value fairly distributed? Are the startups and entrepreneurs who supply the bulk of the innovation and use the Big Tech cloud as a distribution channel, the major beneficiaries of network effects? Given what we know of the way value is distributed across the tech industry - a thousand unicorns representing only a third of the market cap of the Big Five - it certainly doesn't seem that way.

    The benefits that accrue to the largest tech giants are not merely about the cheap capital they have access to, and the acquisition currency their stock enables. When startups deploy their applications to these platforms, they generally have to use the proprietary APIs and platform-specific tools provided by the cloud vendor. Over time, the level of vendor “lock in” increases, as do the costs of leaving the platform. Additionally, the cloud provider has an immense amount of visibility into the traffic each new hosted application attracts, where the traffic comes from and what type of traffic it is. If there is an idea that’s beginning to take off, a cloud vendor can get a good sense of direction on what roadmap additions to make, or which acquisitions to plan.

    In essence, young businesses take grave risks to create new types of applications and functionalities, then – finding that the only way to distribute their product is via a small group of massive cloud providers – they give away critical usage data and create deep tie-ins to the cloud provider’s stack. As the young business continues to scale and grow, the cloud provider earns more and more – for compute, GPUs, network bandwidth and storage – all billed at significant premiums, while their own operating leverage increases and incremental costs shrink. Due to the success and hard work of their clients, the cloud provider develops more leverage with hardware suppliers, reducing prices and expanding margins even further. A virtuous cycle… but only for the big tech cloud provider.

    This is the source of the gravity we spoke of earlier; it is not borne of technical innovation or uniquely differentiated operational excellence. It is mostly a function of large scale capital applied to create advantage.

    “But isn’t this just the way the world works?”, some might ask. Perhaps, but the consequence is that it is much harder for a small company with a compelling product to break out and rival big tech companies, who also have many other ancillary, software, hardware, digital media and online retail businesses. In fact, they can draw upon their reserves in one unrelated business to “dump” data center/cloud capacity at low prices for significant periods of time. In doing so, they onboard startups and entrepreneurs lured by the attractive price of "free". All major cloud vendors engage in this type of "capacity dumping" by offering first-year usage credits or other similar mechanisms. In this way, they undercut smaller hosting providers, attract small businesses to adopt their platform and then encourage deep tie-ins with their proprietary APIs, creating lock-in.

    The increasing financial resources and strengths of the few companies that benefit from “cloud gravity” create an unequal environment in other ways too. For example, due to the immense disparity in scale, a handful of giants can acquire up and coming companies as soon as they become even a remote threat. A billion-dollar acquisition of a photo-oriented social media startup is one example from the past few years. An ultimatum to an enterprise messaging company, “Sell or we’ll undercut your offering and offer something equivalent for free”, is another. Tech giants can fund massive lawsuits compelling others who seek to challenge some of their behaviors; a recent lawsuit between a game developer and a major tech giant running a major app store is but one example. They can also create massive delays in governmental contract awards simply by leveraging the power of their balance sheets to fund lawsuits, coupled with the growing expertise in campaigning for their causes in Washington.

    All these powers are a result of concentration of capital. And the capital accumulates very rapidly for them because they essentially control distribution for the entire industry. They benefit not only from their own work, but are in a position to seek rent from the successes and innovations of all others.

    I am not a fan or proponent of excessive regulation. By highlighting the issues above I am not asking for government intervention or regulation. But before someone occupying the right government office notices that these cloud platforms are benefitting as near-monopolies, perhaps the large cloud providers should think about mending their ways. For example, they could:

    • Agree to limit their shareholding (perhaps as minority investors) in smaller companies that are active in their core business areas.
    • Agree to standardize on a programmatic interface – much like POSIX did for operating systems – so that all clouds can be programmed and provisioned with the same APIs, preventing vendor lock-in via multi-cloud standards.
    • Publish this programmatic interface and its specification as open source so that companies can build their hosted applications with APIs that would seamlessly work on large public clouds, in their own private data centers and on infrastructure supported by smaller hosting companies.

    Constructively oriented and competent governments around the world might also take note of the fact that in the digital economy cloud infrastructure is akin to the highway. It is the mechanism by which innovative digital products are distributed to customers. Providing incentives to smaller companies to encourage them to build their own distribution infrastructure means that a larger number of up and coming tech companies will have an opportunity to enter the top echelon, leading to a more balanced and innovative ecosystem.

    But as I said, the answers don’t lie in seeking government involvement. Nor do I expect the beneficiaries of this disparity to revolve the disparity willfully. What the government or tech giants won't do, the next generation of entrepreneurs might be able to do themselves, with the emerging "Web3" and a move to decentralization.

    First generation decentralized networks did not have the necessary performance to host large scale interactive applications. But new decentralized protocols are resolving these limitations. Newer decentralized networks operate in layers, with foundational layers being slower, highly secure and reliable and associated higher speed layers where much of the work happens. These high speed layers still reconcile data between all layers but they do not require too many interactions with the slower, consensus driven layer. In part, this allows them to support the execution of high speed transactions. Within the crypto space, many such systems are being nurtured. Some projects, from the new Ethereum specifications to Solana and Internet Computer, are attempting to solve performance problems from different angles. It is still just the beginning and many better ideas are around the corner.

    Given the amount of brilliant people now active in the decentralized systems community, it is only a matter of time until the gravity of the centralized cloud – the force that feeds on the youthful innovation and dynamism of young companies – will give way to the centrifugal force of decentralization, enabling individuals, small teams and startups to win a greater share of the value they create. Perhaps even making it so that the scale that results from an entrepreneur’s work won't fund a large company’s efforts to build a rival application that will destroy their own dreams. With decentralization, the underdogs might be able to own more of their success.

    This new digital infrastructure is being built as we speak. One that is inherently fairer and which will prevent systemic flaws that create an unhealthy concentration of capital and power. Might a new dawn rise that unlocks innovation beyond what any of us have imagined so far?



    Read more posts by this author.

    Decentralization Is The Future Of Software