This week we will continue our study of the semiconductor industry by reviewing NVIDIA’s 1st Quarter conference call transcript. The transcript was created by Seeking Alpha.
Disclaimer: The videos below do not belong to us. The creators are mentioned in each video.
Out of the nine analysts on this call, I'm only familiar with Stacy Rasgon and Vivek Arya. Over time, I expect us to know the other analysts since they also focus on the semiconductor industry.
Below is a video with Vivek Arya giving his opinion on NVIDIA's stock price. I kept replaying the beginning of the interview when the commentator laughed at Vivek saying NVIDIA's stock price is cheaper today than yesterday. He probably thought Vivek was losing his mind.
I know I'm late to the party, but I didn't realize that Jensen Huang was the founder of NVIDIA. According to NVIDIA's proxy statement, he owns 3.51% of the company.
Colette Kress, the CFO, joined the company in 2013 and has great experience in the tech industry. Prior to NVIDIA, she was with CISCO, Microsoft, and Texas Instruments.
Professional Visualization Platform: A business unit of NVIDIA that caters to the design, manufacturing, and content creation industries.
Generative AI: Artificial Intelligence that creates texts, images, or other media based on what a person types in. ChatGPT and Stable Diffusion are examples.
NVIDIA's data center revenue of $4.28B is higher than Intel's $3.72B and AMD's $1.3B. NVIDIA's data center revenue is up 14% from Q1 of last year while Intel is down ~39% and AMD is flat. NVIDIA seems to be capturing market share in the data center segment.
Accelerated Computing: A form of parallel computing that separates the data intensive parts of an application so that specialized hardware can process it while the CPU works on the less computing intensive parts of the application.
TCO: Total cost of ownership.
Hopper and Ampere: Both of these GPU's are used in data centers and cloud systems. Hopper is the successor to Ampere. The Hopper model is H100, and it is built using TSMC's 4 nanometer process. Meta, OpenAI, Stability AI (creator of Stable Diffusion) use the H100.
I remember Stacy Rasgon mentioned in the Bloomberg interview that NVIDIA is hard to compete against because they not only provide stellar hardware but also software. NVIDIA provides its customers with software that helps them build AI models faster and cheaper than if the customer had to build the models from scratch.
Pre-trained Models: Instead of building an AI program from scratch, pre-trained models allow developers to use an already built model that can be customized based on the developer's needs.
Framework for data processing and curation: A framework used to create, organize, and maintain data so that it can provide valuable insights.
Systems for Fine Tuning: The process of taking an AI model that has been trained to perform a specific task, and then adjusting the model so it can perform another similar task. This process is cheaper and faster than building a new model from scratch.
Aligning: The process of making the AI perform the task its human operator instructs it to do.
Guardrailing: The process of making sure the conversation between a human and AI stays on track.
Optimized Inference Engines: AI system that applies rules to a database to determine new facts and relationships.
Nemo: Software organizations can use to build their generative AI models.
MLPerf Industry Benchmark: "ML" stands for machine learning and "Perf" stands for Performance. MLPerf is a benchmark that measures the performance of the software, hardware, and services used in AI.
GTC (GPU Technology Conference): A conference created by NVIDIA that focuses on AI, computer graphics, machine learning, data science, and autonomous driving. Data science is the study of data to extract valuable insights.
Software Stack: The different software used to build a new app or project.
Vector Database: A database that use numbers to represent aspects of the data.
G2 Virtual Machine: A digital version of a physical computer.
CONCLUSION
I skipped some words (Proprietary knowledge-based sector database,
L4 Inference) in the transcript because I could not understand the explanations I found. I will do some research and revisit these terms at a future date. God willing, I will see you next week to go over the next 5 pages of the transcript.
Slowly but surely we are becoming experts on the semiconductor industry.
Stay safe, stay blessed, and God willing, I will see you next week.
“Therefore, my dear brothers and sisters, stand firm. Let nothing move you. Always give yourselves fully to the work of the Lord, because you know that your labor in the Lord is not in vain.”
1 Corinthians 15:58 NIV
Comentarios