Meanwhile, a handful of quiet researchers were in the trenches building the real breakthroughs: RDBMS, B-Trees, hash indexes, SQL, and later NoSQL and distributed SQL - along with new hardware architectures that actually made data systems scalable, reliable, and foundational to the global economy.
If someone had bet hundreds of billions on flat files simply because they mistook raw compute demand for actual utility, it would be laughable in hindsight. That’s the feeling I get watching today’s frenzy.
I don’t understand the impatience of supposedly smart tech leaders racing to pour tens of billions into selling and scaling technology that is still unstable, rapidly evolving, and nowhere near its final form (today it is MCP; tomorrow it is Context Engineering; next it is code-execution). Actually, I do understand: FOMO and opportunism.
What I’d rather see is impatience in funding research, fundamental scientific progress, and the innovations that will make AI genuinely more useful, robust, and economically meaningful.
And we can’t ignore the societal cost. People are being laid off, teams dissolved, and careers and lives disrupted - all for technologies that remain expensive (don’t be deceived by the subsidies; every major AI company is burning cash), immature, and unpredictable.
As the old saying goes: those who fly too close to the SUN (Microsystems) eventually get burned.
https://www.aboutamazon.com/news/company-news/amazon-ai-investment-us-federal-agencies