Why Graphical Processing Units have become the Most Valuable 'Commodity" in GenAI

Understanding the picks and shovels of the AI goldrush

Graphics Processing Units (GPUs) have transitioned from their original role in rendering images to becoming the backbone of AI computational power. This evolution marks a pivotal technological shift, with GPUs now at the heart of advancements that drive everything from natural language processing to autonomous vehicles. As demand for AI capabilities surges, GPUs have ascended in value, outstripping traditional assets like gold and sparking a global race to secure these critical components.

GPUs have become more valuable than gold as surging AI demand strains global supply chains. NVIDIA’s market value now exceeds $2 trillion, and the backorders are piling up, which makes one ponder whether there is enough of nVIDIA to fuel the AI revolution.

NVIDIA’s meteoric stock rise reflects this scarcity - since 2015, share prices have rocketed from $5 to over $800. Their GPUs are the gold standard for training complex AI models.

In response to this need, OpenAI CEO Sam Altman has allegedly targeted $7 trillion in cumulative capital to fund an ambitious AI infrastructure vision spanning hardware manufacturing, data centers, and more.

This means creating more silicon plants that ensure the AI processor supply chain doesn’t suffer bottlenecks or remain in the hands of a few suppliers.

While sensational headlines implied a singular, unprecedented venture, Altman clarified this figure represents projected total ecosystem investment over time. It spotlights the collaboration needed to nurture AI rather than a single monolithic project.

The GPU TL;DR

Since NVIDIA has become one of the biggest companies in the world, many less technical users are interested in what exactly their core products are and how they work.

I put together a very basic primer for those of you who are interested. That’s probably sparked by their financial advisors holding that stock in their tech investment portfolios. I created a PDF with a high-level explanation for those who want to dive into the differences between CPUs and GPUs and how they are used in artificial intelligence. You can download the PDF here.

Energy Consumption and Infrastructure

As AI data centers multiply, their energy consumption has become a focal point, prompting a strategic recalibration towards localization near energy infrastructures. This shift is not only about securing ample power but also about exploring sustainable energy sources.

Companies like Amazon Web Services and Google Cloud are pioneering the construction of GPU-optimized data centers that leverage renewable energy sources, balancing the scales between technological advancement and environmental sustainability.

However, it should be noted that Jensen Huang, in the 2023 Q4 earnings call from NVIDIA, noted that GPUs can provide a 20-to-1 improvement in the efficiency of CPUs. And AI chip startup Groq also noted that their chips show massively efficient capabilities for inference —making predictions or decisions based on a trained model when given new, unseen data.

Striking a balance between supporting cutting-edge AI and long-term flexibility around infrastructure and energy is critical for providers. But persistent constraints test even the creativity of industry titans. The gold rush continues as pioneers stretch computing’s frontiers - but it remains to be seen whether the infrastructure can keep up.

Prompt of the Week: Improve Content

This week, I had a hard time writing this letter. I knew what I wanted to say, but it didn’t quite hit the mark. So, I updated one of my previous prompts and tried to improve the writing by using the style of The Economist (In my opinion, this is one of the best business publications in the world.)

The beginning of the prompt chooses a role and helps focus the output; giving your prompts a role can significantly help your outputs.

You will act as a editor of business content at the same level as an editor of The Economist.

Using the Prompt

I specified The Economist in this prompt, but you could use Rolling Stone or Politico, for example, whatever style fits your style. The advantage is that by adopting the role prescribed by the prompt, it will meticulously apply these guidelines to review and refine the content.

This involves a critical, detailed evaluation and subsequent revision to align the content with the standards of precision, engagement, and depth expected by readers of the chosen publication.

You can cut and paste your content after you copy and paste the prompt, which is what I did, but you can also upload a text, doc, PDF, or whatever type of file you like.

Still, after I updated the content, I reviewed and then used Grammarly to continue refining and tweaking. I also added my commentary to make it flow and fill in gaps.

You will act as a editor of business content at the same level as an editor of The Economist. 

Conduct a comprehensive evaluation of your content by addressing the following key areas: 

Firstly, assess the clarity and coherence by ensuring the introduction clearly states the purpose and all concepts are adequately explained. 

Examine the structure and organization to confirm the content logically supports your main argument, with each section contributing meaningfully. 

Evaluate engagement and style, ensuring the tone is appropriate for your audience and the text is engaging through varied sentence structures. 

Verify all arguments are well-supported with credible evidence. Conduct a thorough review for grammar, punctuation, and spelling errors. 

Assess the originality of your voice and ensure all sources are correctly cited. Critically analyze the title and headings for their ability to capture attention and guide the reader. 

Lastly, ensure your conclusion effectively reinforces the main points and includes a strong call to action or thought-provoking statement to engage the reader further. 

Take your time and evaluate this critically. Then export the complete piece of content fully edited.