Disclaimer: ALL the opinions presented in this article are my private ones. They should not be associated with any of my employers: neither current nor past ones. The strategic directions I identify/refer to in this article are based solely on 100% PUBLIC information - I am not leaking any internal strategy plans/discussions/considerations.

A few years ago, I used to do some IT projects for/with banks. It was a stormy period when many banks noticed that: they had spare money to invest, and modern technology offered a bunch of entirely new capabilities banks didn't use so far. They've added 1 to 1 (yup, banks are good in basic maths), and that's how they've decided (with vital aid from "benevolent" consultants) that it's the highest time to elevate the game by:

  • peppering the bank platforms with personalization & social networking (Facebook style ...)
  • introducing "fun" features like gamification, "helpful" push notifications with marketing offers, and such

The reasons were obvious - to show that banks are "cool" (especially when you do banking with friends ...) and to increase so-called clients' engagement (the time spent on the platform, hence the likelihood of buying more banking products).

It did not work, and I've already written about that here. I firmly believe that banks are "doomed" to play the role of low-level "warehouses" of money. And there's nothing "cool" about that. You "like" your bank pretty much to the same degree you like your local water company or electricity provider. Money itself is NOT exciting: it's what you can do with them that could release the happiness hormone into your bloodstream.

What happened next?

That (of course) opened a huge opportunity space for all the companies who would like to build innovative businesses on top of financing services (fully commoditized) provided by banks. There are plenty of examples: e-commerce, which allows installment or deferred payments, collective (group) shopping, or crowd-funding the products that spark crowds' imagination. All these new (higher level) concepts provide extra value (for the end-users) but can only exist with their banking/payment dependencies.

But couldn't it be banks that provide this new, innovative tier of services? It makes a lot of sense: fewer dependencies, fewer parties (who need to get their operational margin), and potentially reduced simplicity. It would also be very beneficial from banks' perspective - unique, innovative services could reach a much higher ROI than commoditized products.

Well, it's technically possible (w/o any doubt), but there are a bunch of real-life constraints that prevent it from happening:

  • Few organizations are outstanding at more than one thing (the focus issue).
  • Banks' complexity and regulatory requirements limit how they operate - and that hamstrings creativity within areas that should not be subject to those constraints.
  • Innovation is all about learning and exploration - for a single company that succeeds in a given field, hundreds of contenders have failed on the way; we may not see that due to survivorship bias.

That's why the most effective model seems to look like this:

  1. Banks offer their services (to other interested entities) in a flexible and easy-to-adapt way, so they are enablers, not sources of limitations.
  2. Small, flexible companies shape new products (each focuses on a single idea/value proposition) on top of bank-powered, stable, mature financial foundations.
  3. Those few of them who succeed (achieve high levels of adoption and reach profitability) grow independently OR get bought out by entities that do have money (like ... banks)

In the end, everyone wins. And the collective rate of progression is maximized.

But the cloud is different! Is it?

Enough about banks; let's switch our consideration back to the cloud. Doesn't the same model apply to the cloud providers?

They've started with simple, low-level (IaaS) services that did replace the on-premise infrastructure (1:1 or close to). Which was a logical move for a bunch of reasons:

  • adoption was simple
  • unmatched flexibility (one could use cloud IaaS infra for whatever they like)
  • after all - it's an excellent way to share the spare infra capacity (that's how the cloud started, didn't it?)

But there are also significant drawbacks - IaaS has commoditized very quickly. And it's damn hard to compete in such a landscape (where products are highly standardized, and your product is very similar to what the competition offers). You try to make a difference mainly with operational excellence - subject to diminishing returns law - and price (a derivative of the former).

That's why cloud providers quickly started traversing paths similar to banks: building more unique services on top of their infrastructure. To be 100% just, they had (& still have) more assets at their disposal.

How so?

First of all: the major cloud providers were already operating humongous online businesses: Amazon - e-commerce, Microsoft - Office 365, and Google - search & ad business. So they turned into assets some pieces of the tech they've created and successfully used internally (e.g., AWS - DynamoDB, S3; Google - Kubernetes). I call it "reverse-dogfooding".

Additionally, they've started offering managed services based on popular Open Source products (like PostgreSQL, Kafka, or ElasticSearch). It's a decent way to monetize what they are really good at - high operational maturity (both process- and tool-wise). As that concept is quite simple to replicate, sooner than later, it has evolved from provisioning unmodified OSS software to running heavily modified versions that, while still fully compatible with the original product on API/SDK/protocol level, take full advantage of the given provider's ecosystem goodies.

These are all great ideas, but - I dare to say - not sufficient to make a significant difference anymore. Why so?

  1. There's a finite set of technologies one can turn into assets (with adequate potential, maturity, general applicability, etc.). Once you're building something from scratch, your advantage of scale practically perishes.
  2. OSS creators defend their products by switching to less liberal licenses (not only because they don't love BigTech - they want to monetize on their own as well).
  3. There's a popular trend of standardizing even higher-level "Cloud Native" protocols (under the CNCF umbrella, but not only) - to increase interoperability and reduce the chances of getting vendor-locked (examples: OCI, OpenTelemetry, OpenFeature, OpenCost, etc.). That allows the commoditization of services beyond simple IaaS.

So unsurprisingly, within the last few years, we got more and more examples of successful cloud services (with unique or at least standing-out value propositions) built by single-product-focused companies, e.g., CocroachDB, RedPanda, Snowflake, or DataBricks. In theory, the big fish could retaliate (e.g., by raising the cost of egress transfer or hiding behind proprietary protocols/standards) - but that would be traversing a path that has already doomed a few former tech behemoths ...

A new direction?

That's why I believe we've already started experiencing the first symptoms of a tectonic shift in the general strategy of major cloud providers (like AWS, Azure, or GCP). Instead of focusing solely on customers who build "traditional" business platforms (non-deep-tech) in the cloud, they will provide all the necessary tools for the companies who could enrich their (BigTech's) cloud service portfolio (preferably: exclusively) with their additions: next-gen databases, message brokers, AI services, analytical tools, ...

This trend has started with the rapid development of mechanisms like AWS Marketplace, but what I mean is an entirely new level of partnerships that span far behind simplified billing, volume discounts, and some joint marketing. What I have on my mind is, e.g.:

  • Access to low-level infrastructure services that were available only internally until now (like specialized storage abstraction layers optimized for minimum, stable latency).
  • Direct integration with core service tooling and provisioning mechanisms (on par with "native" services of the given cloud provider).
  • Dedicated hardware infrastructure components (like cache layers or networking components), which are either too specialized or too complex to use in more generic use cases.

This approach will mark the next step in the evolution of the cloud (as a concept): solidifying platforms of platforms (that run other platforms ...). The ones who'll succeed in building the most open, fair, and developer-friendly ecosystem will gain a significant competitive advantage over the others. Banks failed to build anything remotely similar (in the Financial Services space), but tech giants are a completely different kind of species, I believe ...

Share this post