$美国超微公司(AMD)$  Everything you need to know about $AMD, in 8 bullet points.

1. $AMD's rise over the last decade has been the result of the company betting on chiplets. Chiplets consist in making a chip out of many small chips, instead of making a big one (monolithic) directly like $NVDA does. This is how it has disrupted $INTC and how it is now poised to disrupt $NVDA.

2. We are nearing the "end" of Moore's Law. Making monolithic chips is getting exponentially harder and therefore, chiplets are the way around the end of Moore's Law. They enable much higher yields at similar if not higher levels of performance. $AMD has been working on chiplets for about a decade now, while the rest of the industry is only just catching up.

3. $AMD is just a bunch of people working together to design chips. The company's success is a result of having bet on the right roadmap (chiplets), but also of its excellent organizational properties. Lisa Su is a world class leader and ensures; everyone feels connected to the mission; employees are empowered and accountable; communication is extremely transparent.

4. The market still doesn't quite understand why $AMD bought Xilinx. Essentially, Xilinx is the #1 FPGA (field programmable gate arrays) company in the world. FPGAs enable chips to change shape autonomously and adapt to whatever computational task. They have the potential, over the long term, to deliver much more efficient computing. FPGAs will allow $AMD to deliver AI acceleration in any of their products.

5. $AMD also bought Pensando, which excels at making datacenters stateful. Stateful data centers hold information about their current state of operations. This data can then be used to train AI models, that enable organizations to run data centers at increasingly higher levels of automation. Pensando came with a lot of software engineers too and is essentially going to enable $AMD to provide the environment for their computing engines to do their best work.

6. The combination of the Xilinx and Pensando acquisitions give $AMD a highly differentiated roadmap. For $AMD to bring chiplets to the world, it's had to fine tune its $AMD Infinity Fabric tech, which excels at connecting chiplets. $AMD is now positioned to use this tech to connect any kind of computing engine. This gives $AMD a structural competitive advantage, whereby going forward it will be able to fill specific gaps in the market that no one else will be able to.

7. $AMD is now transferring its chiplet expertise to the GPU market. $NVDA is busy making bigger and powerful chips, defying Moore's Law. $AMD is meanwhile coming from below, with the same strategy that it used to distrupt $INTC. So long as $AMD's organizational properties persist, it's a matter of time before its GPUs are atleast as competitive as $NVDA's. The latter has a strong software moat, but even a moderate gain in GPU marketshare could lead to considerable returns for $AMD shareholders.

8. In the Q3 conference call, Lisa Su said $AMD has made “significant progress” in the Datacenter GPU business, with “significant customer traction” for the next generation MI300 chip. Additionally–and in line with previous guidance–Lisa Su said on the call that AMD Datacenter GPU revenue will be: $400M in Q4 2023, implying a 50% QoQ growth of the Datacenter business; over $2B in FY2024. Q4 earnings will therefore be pivotal for $AMD.