• MISSION+
  • Posts
  • November 2024 — Tech Optimism and A Possible Future of Cloud

November 2024 — Tech Optimism and A Possible Future of Cloud

There’s a tech optimism in the air. Maybe it started with ChatGPT, but the flag was firmly planted when SpaceX plucked the Starship Super Heavy out of the air with chopsticks:

Why is this so important? Because the aim is that the booster can be refuelled and ready to go up again within an hour. Future generations will include this as one of the milestones marking our path towards the stars.

Less publicised, but Bezos’s Blue Origin just did their first booster landing as well. But it’s not just spaceships: Amazon, Google and Oracle are all looking to build nuclear reactors to power data centres, Macron has said over-regulation and under-investment is killing the EU, and China seems to be sprouting competitors to Boston Dynamic’s robots daily.

Accelerate.

A Possible Future of Cloud

The ‘cloud providers have become rent seekers’ argument doesn’t really land with me, but I see the point of view.

But add AI infrastructure assistants to the mix, and maybe there’s something worth thinking about.

There are two challenges in traditional infrastructure setups:

  1. Hardware management - the physical inventory

  2. Effective infra tooling & management - the people & software that makes that hardware useful

The cloud providers like AWS solved both of these - they offered almost unlimited hardware, and sophisticated tooling to let developers provision, setup and manage that hardware.

I don’t suggest any non-specialised company tries hardware management - it is commoditised, undifferentiated heavy lifting. But a niche company could offer bare metal servers and storage at extremely competitive costs.

Once the ‘tin’ is there, a management layer needs to be put on top. The necessary complexity of this layer is proportionate to the complexity of your company. A one-person indie hacker needs almost no complexity. A highly regulated, multi-entity, multi-country organisation needs an extremely sophisticated setup.

In my mind, that is the biggest issue with the “just do it yourself” argument - the skill to implement & manage that layer is non-trivial.

Enter AI infrastructure assistants.

Imagine an assistant where you can ask in natural language, “Provision a multi-site Postgres database within the ‘Asia equities’ virtual private cloud, in the dev environment, with XYZ specifications”. Or “Copy application X infrastructure from environment P to environment Q”.

With this in place, suddenly all that matters is having efficient access to the bare metal. Serverless abstraction layers would just be different API calls - it’s the same basic concept.

The enterprise has been using colo data centers since I joined the workforce more than twenty years ago, so “on prem” hasn’t been literally on premises for a long time. The company that offers the best “new age” interface to that compute and storage may just offer up an alternative to cloud!

Interesting Articles

Mission Accomplished

We’d like to congratulate our client Chocolate Finance for the launch of their very handsome VISA debit card!

It uses the raw VISA FX rate with no markup, so is a great travel card. Sign up here. Well done everyone involved!

Feature A Fractional

We’re On A Mission

Until next month - Ned & the MISSION+ team.

Enjoyed this? Share this with a friend who might too!