• AI Periscope
  • Posts
  • All eyes on Nvidia’s earnings report this week

All eyes on Nvidia’s earnings report this week

And: Nvidia, a peek into operations of a Top 10 supercomputer | How much electricity does AI consume?

Author

AI Periscope
February 19, 2024

Exploring below the surface of AI headlines.

Summaries | Insights | Points of Views

In Today’s Edition

Nvidia’s earnings on Wednesday

Summary - Investors are eagerly awaiting NVIDIA's earnings report scheduled for after the bell on Wednesday, February 21st, as it is expected to impact the market's view on the 2024 outlook and the economic influence of AI. NVIDIA's stock has soared by 46% since the start of the year due to AI-driven optimism, helping to push the S&P500 to all-time highs. As a key player in AI technology, and recently surpassing Amazon in market cap, NVIDIA's earnings report could be a bellwether for the industry in the near term.

Buoy points:

  • NVIDIA's shares have surged by more than 46% since January 1.

  • The company's market cap growth has outpaced chip maker competitors like Intel.

  • NVIDIA's performance has helped propel the S&P 500's to all-time highs this month.

  • The street expects a significant increase in NVIDIA's quarterly earnings and revenue.

  • Options traders are anticipating a large stock price movement post-earnings report.

  • Positive earnings could boost AI optimism and support the current market rally.

  • A report that merely meets expectations could lead to a sell-off due to high investor anticipation.

POV - Nvidia makes up approximately 3% of the S&P500. Investors will be listening carefully to the earnings guidance on the call after the bell on Wednesday. Earnings guidance is key to moving future price. With some street analysts’ recent price targets as much as $1200, the exuberance this week needs to be matched by the guidance. With all the recent big developments in AI, Sam Altman’s unprecedented $7T funding desires, and Nvidia’s 85% grip on the AI chip market, will Nvidia push the market another leg up?

AI Supercomputer

Summary - NVIDIA's Eos is a top 10 supercomputer and is setting the stage for the next wave of AI innovation. With its advanced DGX AI architecture, Eos is designed to tackle the most demanding workloads, from training large language models to powering quantum simulations. This supercomputer combines NVIDIA's cutting-edge technology, including 576 DGX H100 systems and Quantum-2 InfiniBand networking, to deliver an unprecedented 18.4 exaflops of FP8 AI performance. Eos can serve as the accelerator for enterprises aiming to scale their AI innovations.

Buoy points:

  • Eos is powered by 576 NVIDIA DGX H100 systems, a total of 4,608 H100 GPUs, making it a behemoth in AI computing.

  • It ranks No. 9 in the TOP500 list of the world's fastest supercomputers.

  • Optimized for ultra-low-latency and high-throughput demanded by large-scale AI computations.

  • Eos's network architecture supports data transfer speeds of up to 400Gb/s.

  • NVIDIA's full-stack approach integrates accelerated computing, networking, and AI software, providing a comprehensive solution for AI development and deployment.

  • The DGX SuperPOD architecture at Eos's core represents NVIDIA's commitment to advancing AI technology and infrastructure.

  • Eos is described as an AI factory, emphasizing its role in enabling the rapid development and scaling of AI models and applications.

POV - An interesting peek inside the engine of AI today. Aside from its valuation and the parabolic move of its stock price recently, I find it hard not to be enamored with Nvidia for the sheer story of American innovation and success story that it represents.

AI power consumption

Summary - AI's electricity consumption is a widely discussed and somewhat mysterious issue, largely because major tech companies have been less than transparent about it. The energy required to fuel AI models, whether for training large language models or for everyday uses such as chatbots and image generators, is significant but difficult to measure accurately. The contrast in power usage between training these models and using them in real-world applications underscores a major environmental impact, leading to ongoing discussions about sustainability in the tech sector.

Buoy points:

  • AI's energy consumption is significant but difficult to quantify due to variable configurations and company secrecy.

  • Training a large language model like GPT-3 is estimated to use just under 1,300 megawatt hours (MWh), about as much power as consumed annually by 130 US homes.

  • The energy usage trend of AI models is upward, due to their increasing size and complexity. And, OpenAI is on the verge of GPT-5.

  • At the individual level, image generation uses 2.907 kWh per 1,000 inferences, equivalent to the energy used for charging the average smartphone 242 times.

  • By 2027, the AI sector could consume between 85 to 134 terawatt hours (TWh) each year, which is roughly equivalent to the annual energy demand of the Netherlands.

POV - There are continuing reports of big tech companies increasing investments in renewable energy. For example, Amazon has 500+ renewable energy projects globally. Others like Microsoft are reportedly considering nuclear power. The increasing energy demands due to AI and cloud computing, along with an aging power grid, will surely require new investments in energy sources, in renewables, nuclear and otherwise. Will we be able to keep up?

Don’t forget to check out other AI headlines in The Ocean.

The Ocean is designed to help you stay informed with the wide array of AI headlines, giving you just the snippet of news that you need if you are short on time, but also gives you one-click ease if you want to dive deeper.