"Reasonable profits are profits that let us keep doing research, keep our services running, and give both our employees and our shareholders a return. If our prices ever get so low that we have to operate at a loss, that won't last either."
Then Han Luyi took the initiative to change the subject and explain his ideas for building a domestic ecosystem.
"We've now established a working vertical pipeline, from the chips at the bottom to the model applications at the top. But if we want our domestic peers to join in, we need to open up that middle layer. Right now, everyone is constrained by Nvidia's ecosystem, not only because CUDA's operators are easy to use, but also because many of their implementations are protected by patents. Even if we wanted to use them, we couldn't."
"But I intend to release the Intermediate Abstraction Layer we've already implemented for free, without patent restrictions. That way, as long as chip manufacturers can adapt their chips to the operators in the abstraction layer, any model developer will be able to use those chips."
"Of course, that adaptation is a huge undertaking, and we're also willing to send technical teams to help other companies tackle it."
Elder Wang shook his head. "Boss Han, that would practically make you a saint. What do you gain by helping your competitors solve their technical problems?"
Han Luyi smiled. "There are benefits, actually. Our technical resources are limited, and we can't help everyone at once. So my idea is that whoever gives Source Intelligence priority access to their production capacity gets priority technical support from us."
Ever since Glutinous Rice Ball's launch, computing resources had been stretched thin. User numbers had skyrocketed, and there wasn't even enough computing power for inference.
To begin the next round