
There was big AI news: OpenAI signed a major chip deal with Broadcom.
When people talk about “AI chips,” they usually mean two kinds:
chips used to train AI, and
chips used to deliver AI to users (this is called inference)
NVIDIA is the clear leader in training chips. But inference is a little different. For this part, speed, cost, and how well the chip fits a specific purpose matter a lot. Because of this, custom chips—made for one special task—can be faster and cheaper than NVIDIA’s general-purpose chips. OpenAI’s deal with Broadcom is about building these custom chips at scale.
This idea isn’t new. In the early days of the internet, Amazon and Google designed silicons that worked perfectly with their own software. Apple has done the same with the iPhone—building chips inside the company so the hardware and software fit each other exactly. The best performance always comes from reducing the “gap” between hardware and software.
AI will still rely heavily on NVIDIA’s chips, but more companies are now trying to build chips that match their own needs.
Clothes made for everyone sell easily, but they will never fit as well as clothes made just for one person. The quality and satisfaction of something custom-made are hard to beat.
Investment works the same way. We often hear stories like “someone made a lot doing this” or “this strategy worked for them.” But these moments are usually just the brightest highlights—one screenshot from their whole life. We don’t see their full portfolio, the work behind it, or the losses they took along the way. Copying them almost never works. In my ten-plus years of investing, I’ve never seen imitation lead to real, lasting success.
The lesson was simple: strategies that work for others won’t necessarily work for me. Only the strategy that fits me can last long and increase my chances of long-term success.
So I need to find the clothes that fit my own body. And to know what fits me, I need to understand myself well. A strong body needs a strong mind. And a strong mind helps build a strong body. With both, I can finally wear my own custom-made “armor.”
Just as AI companies add custom chips on top of general ones to get the best performance, I need to build strategies that fit who I am—my life, my work, and my investments. This requires thinking constantly, training my mind and body, and adjusting my approach again and again.
Good software and good hardware.
And the alignment between them.
Building my own software—and finding the hardware that fits it—is the path I want to walk in my investing and in my life.
Steve Lee
No comments yet