What to do with infinite compute power?

Selvaraaju Murugesan
2 min readApr 12, 2021

--

Proliferation of cloud has democratised access to compute resources and services for every enterprise. Most of software are delivered via web browser, on-prem data centres are becoming obsolete, and many services are commoditised for easier use.

Because of cloud, we have access to infinite compute power. For reasonable prices, even a common man can buy cheap compute power at scale. When compute power was scarce, we were unaware of solving big problems using big compute engine.

As tech grows, big computational intensive algorithms are cheaper to execute leading to many big problems being solved. For e.g problems in climate change, simulating big bang, cracking DNA code, vaccine design, etc.. are being solved using big computational engine.

For cloud providers, this means only a few big customers / companies are using vast amount of compute power to solve complex business problems to either build new products/services, maximise user/product experience, etc…

Based on empirical observations, the compute power have grown exponentially while the big problems (that utilises large amount of computer power) seem to grow linearly.

Even though a common man can buy cheap compute at scale, they do not have raw material aka data to power those compute !

Many companies in future will open up their data repository to spur innovation. Many people in future will have access to unlimited compute power, reasonable access to big data and also will have skills to solve big problems! Since the purpose of the compute engine is to help solve business problems.

History has shown that every solution to current business problem will create a new business problem. eg. to move from one place to another quicker accelerated innovation from horse to cars to aeroplanes. However this leads to various other business problems in safety, pollution, and so on.

I believe that big compute power can be used to focus on solving humanity problems at scale ! Thoughts !

--

--