SaaSpocalypse? Starbucks is building its own software now

On 9 July it emerged: Starbucks is developing its own software with generative AI to replace systems from Microsoft (inventory tracking) and IBM (maintenance management). The backdrop: around USD 400M in annual software spend, under review as part of a USD 2B savings programme. First in-house builds are due to go live by the end of next year.

A coffee company — not a tech company — believes it can build enterprise software from two of the world’s largest IT vendors itself. Two years ago that would have been an absurd headline.

Our take

This is exactly the scenario we sketched at our conference two weeks earlier as the “SaaSpocalypse”: when creating software becomes dramatically cheaper and faster, the SaaS business model comes under pressure. The three theses from 25 June apply directly to the Starbucks case:

• SaaS has to get cheaper. When the customer can build the alternative themselves, bargaining power shifts. Use that at your next renewal.
• Only real moats survive. Software with exclusive data access or network effects stays valuable. Generic workflow tools — inventory lists, maintenance tickets — are first in line.
• Watch out for lock-in manoeuvres. SaaS vendors under pressure will try to bind customers more tightly. If you’re signing long-term contracts now, double-check the exit clauses.

But “build” doesn’t mean “one prompt and done”

Scaling, quality, reliability and maintenance remain real engineering tasks. The right question is no longer “make or buy?” but: where is software strategic enough to own — and where do we keep renting?

Weighing build vs. buy for a specific system? Reach out to the contact below.