The scalability of > 10x cheaper dev costs

We're winding down our KL trip and will be leaving for Manila tonight. Currently working at VCR, which goes to show that hipster cafes really do scale across the world.

One surprising finding about developer salaries across the cities we've been. Developers, with 2-3 years of experience make:

  • $700 USD/month for Vietnam
  • $3,000 USD/month for Malaysia
  • $10,000 USD/month for US

    In summary, labor in Vietnam is 4x cheaper than Malaysia, and 14x cheaper than the US. There are two conclusions I draw from this data.

    1)
    There is value to be captured between the cost-effective dev costs in SE Asia and AI trainers needed. For example, assume we have two AI scheduling assistant startups, say X.ai and Mimetic.ai. Both start with the same algorithm, the same amount of funding, and the same number of initial users. X.ai is based out of NYC and uses USA trainers and Mimetic based out of Singapore, and uses SE Asia trainers.

    For every funding $, Mimetic is able to hire 14x more training effort.

    Let me repeat that - 14x. So as long as Mimetic scales out its customer acquisition and data input, it can take 14x more the input and train its AI. And in the AI game, its about this speed of training that will enable one learning algorithm to progress faster than the other.

    2)
    There is value to be captured between the cost-effective dev costs in SE Asia and the increasing demand in the US and UK. Two models that can leverage this are dev agencies/shops and recruitment agencies.

    Granted - this is an old school model, but startups like Freelancer or Gigster is an update of that model.