Every technology startup faces the question of whether to focus on a core technology piece vs. a full-stack solution. NovuMind is bringing a full-stack AI solution set to our customers.
We’re in a period where AI is suddenly a top priority for every industry. It’s an exciting time. But unless your company is a Facebook or Google, with in-house AI expertise, how will you seize the new opportunities brought by AI? How will you keep up with your competition, who may be embedding the powers of perception and decision-making into their upcoming products? We will be happy to sell you our NovuTensor chip for these applications. But we know that you probably need more from us than just the chip. You are looking for a complete solution.
This is the essence of “whole product” strategy that the Silicon Valley marketing guru Bill Davidow made famous. When he was at Intel, this meant providing more than just the microprocessor. Customers needed development tools, training, and an ecosystem of supporting technology. Years later, NVIDIA followed a similar strategy, building out the CUDA ecosystem to accelerate GPU adoption.
Making AI easy to use is part of our vision. We want to empower all kinds of companies, in all industries, to use AI to improve their products. In turn, AI will improve the lives of people around the world.
But full-stack means more than just making the customer’s job easy. It’s also about quality and performance. Selecting, training and tuning deep learning models is somewhat of a black art. Popular models have not been designed with industrial/embedded applications in mind. We can provide a combination model and accelerator chip solution that is optimized for the application and environment. This is the idea behind our plug-and-play NovuBrain solutions. Using our NovuStar supercomputer for model training can also make a huge difference. Let’s say the customer has a big data set and training their model in-house used to take 4 days, and we can reduce that time to 1 hour. Now they can do 100 times more iterations than before. This will result in a better performing model.
Building a full-stack AI company has its challenges. More projects to manage, more competencies where we need to hire experts, more areas where we need to invest. But it gives us a stronger position in the market and it brings lots of benefits to our customers. It is an essential part of our strategy to democratize AI and bring all the benefits of AI into peoples’ lives.