#3: Cloud Computing as a Modern Utility
Also in this edition: LatAm as the next frontier venture market after Asia?
For my first 2 posts, I’ve been doing explainer essays. To be honest, one of the reasons I started a substack is for me to crystallize what I’ve been reading either on the news, on twitter, or from other newsletters. I haven’t settled on a format yet, so bear with me as I continue to experiment.
Cloud Computing is the New Electricity
Byrne Hobart’s The Diff is one of my favorite newsletters, even though I don’t exactly have the money for a subscription. In “Electrocloud!”, he makes the case for cloud computing as a new, modern utilities business.
We all know that Edison is credited with inventing the electric bulb, but apparently Edison wasn’t the best businessman; he charged a flat fee for electricity while he sold continuously improving lightbulbs which meant that clients would replace them less often, meaning less income for Edison.
His assistant Samuel Insull eventually ran his own business, and after a trip to England he found a better pricing model: usage-based pricing. This had 2 benefits:
Electric companies can just focus on supplying electricity instead of having to also sell lightbulbs.
Electric companies don’t have to care about lightbulb improvements. They just need to care about getting more usage.
The early version of home appliances was born out of electrification, first by plugging them into light sockets. Or as Hobart put it, they “were a primitive instance of leveraging undocumented APIs: a socket doesn't know what's in it.” Even at that time, the familiar concept of utilizing resources at non-peak times existed. Appliances were used during the day since lightbulbs weren’t needed during the day.
More importantly, electrification completely revolutionizes manufacturers. Traditionally, they had to invest in a hydropower or steam-powered generator, which meant that their manufacturing capability is limited by their power generation, and they had to invest in capex to raise the limits of their generators.
Electrification allowed manufacturers to have the option to use electricity from a centrally-generated power instead of building their own power generators. Meaning, they can treat electricity as an opex instead of capex, which means less cost and time needed to start utilizing electricity.
When you compare the above with cloud computing, electric power generators = servers. Companies built in-house mainframes for computing, but as the limits of computing power starts to improve (Moore’s Law), data centers start to become the norm. With further computing power improvements, now cloud computing is and will become the norm in the near future until our next technological breakthrough… maybe edge computing brought about by even more powerful microchips and 5G?
The real gem in Hobart’s writing is the observation that not only do companies save on time and infrastructure costs, they also save on invisible organizational costs. Imagine starting a SaaS company, but instead of using AWS/GCP/Azure, you have to build your own server rack before you can push anything online, and hire an in-house team to manage the server load or physically build more servers when traffic surpasses expectations.
Using current cloud solutions instantly reduces the execution risks of new startups, since there is less that can go wrong as startups scale. Instead of spending so much time on optimizing traffic and reducing data, you can just pay to use more computing power and keep your website/SaaS/etc up.
Hobart argues that insane startup valuations might have some to do with this de-risking due to relative certainty of infrastructure costs as startups scale. This allows startups to raise money based on their TAM and long-term margins, making future unit economics prediction much more accurate than before.
LatAm: US biz model, local execution, Asian lessons?
Kevin Xu had a piece on Mexico’s startup scene and LatAm’s as an extension, the essence of which is similar to other developing countries: US business model, local execution.
What’s really interesting though, is Mexico’s similarity with Indonesia or Vietnam:
As they find product market fit, they start to look to Asian tech companies for inspiration while morphing them to suit local needs.
The Colombian unicorn Rappi started as a grocery delivery app, but they’ve followed the super app playbook of Meituan and Grab to become a regional giant in LatAm.
He also followed up with news of the first Mexican unicorn, Kavak, a marketplace for used-car who has their own auto-financing program. It’s not improbable that they’ve taken lessons from the used-car platform war that's already wrapped up in China. What’s peculiar about this is that the Southeast Asian used-car platforms Carro and Carsome are exactly playing the same game, only in different regions.