In my experience, simply setting your sails in the right direction is not how you find success. I have found that when luck and determination meet is where the magic happens
The journey begins by first daring to take the plunge, leaving the shore without knowing which direction you are heading towards and why. That is what it was for me when I left BNR (my employer). The timing was not “right” either, as my husband was still struggling with his own startup. Leaving my full-time job to start my own company would mean that our income would be one-fourth of what it was when we were both employed. And we would be making that sacrifice for several years.
I did not realize then that my company would have the benefit of the Internet tailwinds to push me in the right direction.
I was trained as a hardware engineer, which came very handy in starting my company. Early in my career, I saw Intel 8080 introduced, the first 8-bit microprocessor, and software engineering gradually got introduced. I remember a suited Intel sales guy walking into the company, and engineers looking for a break from their daily grind welcomed his presence. He handed each of us the specification book on this new wonder chip and gave his sales pitch. To us it was exciting to use the microprocessors, as engineers love new technology even if it does not make economic sense. As technocrats, we had little understanding of markets and the economy.
Hardware engineers gradually started to learn assembly language–the glue between hardware and software.
It was then that the tide turned. Software companies, with their minimal Cap-ex requirements –just a PC–exploded. Compared to hardware product companies, the margins of software product companies soared, venture capital flooded in, and startups proliferated. But competition also exploded. The barrier to entry was low: a geek in their bedroom with a computer is all it took.
Today, in the current AI era, the hardware has once again become the centerpiece. Technology has come full circle. The demand is for processing speed and power-efficient GPU chips for GPU data centers or maybe quantum computing in the future. Artificial Intelligence-trained Large Language Models (LLMs) can now code, reducing the value of software engineering. I could have never imagined that being possible when the first microprocessor came out many decades earlier.
The Cap-Ex, advanced GPUs, and massive data centers required for training these models are now in the billions. The software pyramid has become inverted. The traditional VCs’ business model can no longer support this level of capital requirements, putting them outside in building foundational models in the AI game. Only companies like SoftBank and Andreessen-Horowitz can participate. Or, as is the new brand trend, the MEGA-Techs. The traditional startup echo system of VCs funding an engineer with an idea has vanished for now.
Startups, if they use third-party or open source foundational models, will no longer have product differentiation, lowering the barrier to entry. They will not be easily able to develop their own secret sauce.
However, this isn’t a moment for despair. We’ve seen this cycle before, from hardware to software and now back to hardware, but with a crucial difference: AI is the catalyst, the force that reshapes everything in its path. Those who understand the interplay between hardware and software — which is my experience—can navigate the complexities of AI development and deployment and will be the architects of the future.
Many more cycles are yet to come, but we have been creative before. With a culture of resiliency, Silicon Valley will continue to evolve to continue to generate economic gains through technology.