The Whole Stack Wars are coming
The future of software is plastic — this may be how the story arc of the Software Age ends.
Same. I’ve begun referring to this concept as “plastic software”.
Software has always been designed top-down – customer research findings were handed to engineers, PMs and designers, who (even when they run the shortest possible sprints) aren’t observing user behaviour and inferring customer needs in real time. And when they write software, it has to be for all users, or the largest possible subset of all users. Never any individual user (with the exception of those who code for Elon and Elon alone).
Soon a significant proportion of software may begin studying your use patterns live, as you use it, and may begin changing as it infers your needs from how you use it.
You and I might start out with the same features and UX when we register to use a future version of an office productivity suite but your version and mine may gradually begin to diverge.
The features you never use may disappear from the UI, features you rarely use may migrate to lower-priority places, and features the software thinks you might need may be introduced to you.
Features that previously only would have been available in another product you weren’t paying for (such as CRM or social content planning) may begin to arise in your UI as the plastic software interprets your needs.
As the cost of LLM processing continues to plunge the fractional cost of having an LLM write this bespoke functionality will be so small compared to the value of further locking you in to a single software stack that software vendors may well fall over themselves to encourage this to happen.
If that begins to happen, the Boxed Software Age distinctions between word processor, spreadsheet, email, database and design apps could blur and disappear.
At first it might seem you’re switching mode but still within the same app. Then, when the software judges you’re ready for it conceptually, maybe you’ll just be in a single ‘plastic’ app, just called Microsoft/Meta/Google/Apple/Nvidia/Meta/Canva/Salesforce Business, or similar.
When your software infers that your productivity is flagging and you need a rest, it might fold in a few quick, personalised snippets of TikTok-style entertainment, right there in your plastic business app. All the better to keep you in-stack 24x7.
This may facilitate the next stage, the battle to achieve and retain “Stack Capture” of every individual user.
When you can buy plastic software that can serve any of your business or personal software needs, the battle for the major software vendors may be to capture your ‘stack’ – your time in your plastic app, plus your single cloud data storage. All your own content and memories, everything you subscribe to, all your interactions with your work files, and all your personal and work interactions with others.
You will never be logged out again. (unless you get hacked).
Being a whole stack customer could reduce their cost of providing you with plastic software and make you much less likely to ever churn.
We’ll probably want to be whole stack customers because it should be much cheaper for us and our software will work much better.
Employers may try to choose a software vendor which will be the closest match for their ideal employee profile. Employees will surely choose employers who are already on the same stack, or if they aren’t already, go freelance or gig so they can stay ‘on stack’.
One thing I’m uncertain about is whether the vendors which have so far chosen to invest in their own hardware devices (eg Apple all the way, Microsoft largely, Google kinda-sorta) have an advantage in the stack wars over the so far hardware-agnostic stack vendors (Canva, Salesforce, Meta, etc).
Could Nvidia go downstream and initially partner and then go it alone with its own edge-LLM laptops, phones and tablets? Does OpenAI acquire Nvidia or vice versa? Don’t know. Depends how much owning the hardware delivery of UI matters in the future whole stack wars. If “product polish” matters enough to offset the capital outlay of manufacturing hardware.
What happens to the mostly-hardware vendors like Huawei and Samsung or the enterprise-facing-only vendors like IBM? Partner-and-be-acquired, or die.
The vendor battle will be to attract, retain and protect your whole stack value. We’ll cease referring to them as software vendors because that will only describe a small part of what they are to us.
A better word might be “tribe” or “community” or maybe even “nation”.
