AMD’s AI Secret Sauce: Dogfooding

This week AMD had a major AI announcement that was supported by vendors ranging from the leading server, PC and networking vendors, to Microsoft. But the competition in this space is fierce. NVIDIA is the dominant player, but AMD comes with three unique advantages. First, it is a huge advocate of OpenAi which has been a massive disruptor in the software technology space. Second, it has one of the strongest partnering records in the industry and its custom business is arguably the strongest (having captured two of the three major gaming systems). Finally, AMD’s CEO, Lisa Su, is one of the few who was trained to be a CEO. Lisa Su has been recognized as being one of the strongest leaders in any industry.

These advantages are well known. What isn’t well known is how aggressively AMD has implemented AI internally which has allowed it to massively accelerate its own development process and sales efficiency, and to more accurately forecast its future performance.  

Let’s talk about this secret sauce and how AMD is showcasing that in the AI space. It isn’t just what you build but how what you are building helps you become more effective and competitive going forward in terms of AI. 

The Importance of Dogfooding

One of the terms I learned early in my career was “dogfooding.” This is where a vendor who builds a product can prioritize its use internally. Over the years, I’ve worked for and with companies that didn’t aggressively use their own products and then had to defend this lack of use to customers after telling them that the related products were game changers and would pay for themselves. 

When challenged, one CEO I worked for said something to the effect that he could choose whether to supply the technology internally or sell it to customers. This didn’t play well because companies can do both, and there was no reason why a company building a tool that was presented as delivering massive financial benefits couldn’t justify using it internally at cost.

In addition, when a tool is used internally, people have a personal interest in addressing problems that adversely impact their use which can accelerate fixes that otherwise would have to wait for field reports to make it through to development and then be put into a correction plan. When your people are having issues, the speed at which a problem is identified and addressed is much faster.  

AMD is dogfooding its AI technology. Assuming it works (and based on the partners at the launch event, it truly works), this will massively speed up AMD’s development and increase staff productivity with significant potential benefits to the company’s bottom line. 

Applied AI Benefits

One of the bad experiences I had as an internal auditor at IBM was fixing the forecasting process to create an annual 20% inaccuracy only to find that some idiot had applied the 20% correction to the now accurate number. This resulted in the firing of a CFO who I personally liked. Applying AI to financial performance, if done right, not only should shorten the effort required to make that forecast but increase its accuracy by eliminating the human decisions and potential related mistakes inherent in existing processes. So, you get both higher accuracy and a smaller chance that the CFO will get into trouble by significantly increasing forecast accuracy. The stock market really values forecast accuracy, so this creates a significant potential benefit to AMD’s valuation.

In terms of sales performance, IBM’s tests with Watson have indicated significant improvements in close rates and customer satisfaction, suggesting that AMD’s use of this technology in that process should allow it to better address customer needs, more quickly address customer complaints, and better close new business. This too should have a significant impact on AMD’s bottom line and performance in the long term.  

Wrapping Up:

Microsoft is famous for dogfooding, and it has significantly improved its overall market success, product quality and customer satisfaction. The implementation of its AI Copilot feature across the product set and internally should make Microsoft even more powerful going forward. AMD is following that same path with AI (and Microsoft is helping), suggesting that as successful as it is, it will be even more successful going forward.  

AI has that potential. It’s a game changer, and AMD will become (and already is, in a way) a powerful showcase of how effective AI is in improving overall performance and success.