This article looks further into where enterprise software is headed arguing that only if we reuse and share code and make this type of sharing work across stacks and framework we will be able to control the problem at hand - being 500k, or even more, developers short by 2020.
Let's recap what we have discussed so far - in the first article, we have touched upon the evolution of enterprise software, depicting its transition from on-premise monoliths to highly scalable API first microservice architectures - the digital aqueducts and railways enterprise software is built upon today. Predictions from just five years ago that stated we would see one dominant player per software category have not materialized. Rather, businesses increasingly buy specialized systems that are kept in sync through API connections and adapt to the business workflow wherever possible. We’ve also briefly mentioned on the war-stories of companies trying to build one-size-fits-all solutions, as exemplified by Lidl and Haribo. As we have shown, the main problem with this type of implementation is that a growing system size is really hard to control with the organisational structures current IT department can manage.
And in the second article we discussed the current underlying technical revolutions that will drive this necessary change towards more process and human-centred (i.e. customized) solutions - the very fabric modern enterprise software is made of. We’ve shown that the advancing modularity in frameworks and in API design will allow for a new level of code-reuse and a combination of different stacks and styles in an application. Likewise, evolutions in seemingly small things like the way we document code can make a huge impact on productivity, while modern hosting infrastructure is to eliminate a lot of the manual configuration hassle that developers currently face. Thus, current trends in software development will allow for a lot more reusability of code and require significantly less manual configuration via automated deployment, especially enabled through the current Kubernetes trend. This, in turn, will help developers to understand, digest and reuse code much faster than ever before.
Now, let’s enrich our learnings with a few additional facts: the pressure on digitization in firms is enormous. According to the Worldwide Semiannual Digital Transformation Spending Guide of the International Data Corporation (IDC) 2018 (1) global spending on digitisation projects amounted to 1 Trillion dollar and is supposed to double until 2022. According to the latest research by 7-Miles advisors, demand in customized enterprise software will increase at a compound annual growth rate (CAGR) of 33%-43%% over the next 5 years, and amount to 500bn dollars in 2022 (2). Yet, the landscape is scattered, and custom development as well as the alteration of existing systems is done by an extensive amount of agencies, corporate developer teams as well as freelance developers all over the world. A core problem here is that the growth in talent is by no means keeping track with the growth in demand. According to a study by Deloitte, we will be 500k developers short by 2020, which might be a rather low estimate: In a keynote during the opening of their Berlin office in January, Google CEO Sundar Pichai put this number at 1m.
All of this won’t be without consequences: First, these developments will put enormous pressure on HR. We’re already seeing a 15% wage increase in developers’ salaries, which is way above the general market trend. The Bureau of labor statistics of the US gov has some good numbers around that. Second, these trends will also create great investment opportunities for those that make development more easy, automated and human-centered. Every single day that can be saved in an enterprise level development cycle is worth a lot of money, namely between 5 to 10k for a team of developers that include QA, DevOps, PM and infrastructure. This renders it an easy sell for tools that can actually speed up the development process. In light of the developments described above, we assume that this will even increase in the near future: With more capital flooding the market, we will witness a significant acceleration in the evolution of these services and technologies.
Interestingly enough, this market will probably not be conquered by legacy players, as cheaper custom software is not really in line with their code ownership, under which stakeholders are bidding heavily on exploitation of the advantage of closed and proprietary code. As pointed out earlier, Microsoft’s recent acquisition of Github poses an interesting reflex in this regard. Even if parts of the business will always push for expansions in these areas, there remains a fundamental nudge of the core business to make sure new developments do not mingle with the lucrative core business - just look at the way Salesforce forces you to tie your services into its suite of tools, or at least their license monetisation. All of this makes the market very prone to be conquered by innovative and fresh teams.
For now - and intentionally without proposing a solution to this set of problems for the time being - we make a somewhat daring projection. Against the background of the above-mentioned high level-trends of growing demand, shrinking supply and skyrocketing capital inflow, we predict that in 5 years, 5 computer science students will be able to code a complete legacy ERP system in just 5 weeks from scratch. This system will be fully functional and adapted to the exact workflow the business requires. We do not have the slightest idea what implications that may have, but it will surely shake the foundation that most legacy enterprise software systems are built upon today.
The consequences are fairly straightforward. The further we can lower the price point of custom solutions, the more directly we will attack the core market of legacy players. With decreasing prices in custom software development, we will see pressure on the price point of proprietary systems. Custom development teams will prolong their monetisation cycle by shifting from making money merely by development to licensing and maintaining custom enterprise software in the long-run. The lower costs in production will probably be handed over to customers, as the competitive pressure to win a pitch will decrease prices.
This reduction in prices for custom software will slowly, but eventually change the buying behaviour of IT departments. Eventually - yet arguably some 5 years later than in regions more prone to innovation - this will even materialize in Germany, where corporate culture might be at clash with DevOps (article in German). That is especially relevant at a time when the rest of the world is just about to get rid of DevOps altogether.
In general, IT departments will at some point figure out that lifetime costs of proprietary monoliths are not only quite high (due to their well-known particularities), but can also only be adapted to a certain extend to the actual business processes. Contrarily, customized systems offer a number of striking advantages that include lower dependencies and higher flexibility - and the termination of lock-in effects. Will this have influence on the classical one size fits all CRM systems in the standard use-case? This will probably not be the case, yet it is highly relevant for more complex customizations.
This evolution of making possible the creation of standard-components that developers share and reuse will also decentralise the development of these components. Where current proprietary software houses have ten times more salespeople than actual developers, we will see small engineering teams taking over implementations by throwing together small, component based software solutions. In a nutshell: What WordPress did to website themes could also happen to enterprise software development components.
Stay tuned for the next article “Speak flats, cars and offices: How software moves into the shared economy, too” where we will discuss this further.