To The Moon

On Wednesday (25 Sep), we saw $NVIDIA Corp(NVDA)$ $Advanced Micro Devices(AMD)$ and other chip stocks gained, rebounding from recent losses at the beginning of the week.

Even with Wednesday’s gains, many chip stocks have not completely recovered from losses earlier in September.

As tightening trade restrictions and investors rotating to small-cap stocks because they anticipate benefit from the rate cuts that is set to happen on 23 Sep, September have give some hard times for chip stocks.

The rise of the chip makers stocks like Nvidia, AMD, $ARM Holdings Ltd(ARM)$ which rebounded from recent losses was due to the surging demand for artificial intelligence (AI) is likely nowhere near its peak. 

September Tough Month for Chip Stocks

Semiconductor stocks took a hit this month amid concerns about tightening trade restrictions and as investors rotated into small-cap stocks, expecting they could benefit from Federal Reserve rate cuts.

But as we saw that the recent pressure on chip stocks was short-lived as the broader rally in chip stocks this year makes a comeback on strong fundamentals and surging demand for AI. 

I am expecting chip stocks to go into the October month with a comeback strong rebound.

‘AI Tidal Wave’ Going To Trigger A Start of the Tech Bull Run

With interest rates coming down and AI tidal wave of spending expected to rise, we are going to see the start of a tech bull run.

Companies and governments are expected to spend a combined over $1 trillion over the next few years to fuel what they called the "AI Revolution." 

So the key beneficiaries of this surge in AI spending would be chipmakers, especially those who are more into data center and AI end markets like Nvidia and AMD.

AI Inference Competition Heats Up

Nvidia GPUs has dominate the AI training, this is rather undisputed, but there are early signs that for AI inference, Nvidia is well placed in the competition especially in terms of power efficiency. This came from the sheer performance of Nvidia’s new Blackwell chip which seem hard to beat.

Note that Nvidia production of Blackwell is well underway. We could be expecting it soon in Q4 2024.

In the recent AI inferencing competition (ML Perf Inference v4.1.) results released by ML Commons. The round included first-time submissions from teams using AMD Instinct accelerators, the latest Google Trillium accelerators, chips from Toronto-based startup UntetherAI, as well as a first trial for Nvidia’s new Blackwell chip. Two other companies, Cerebras and FuriosaAI, announced new inference chips but did not submit to MLPerf.

MLPerf has many categories and subcategories. The one that saw the biggest number of submissions was the “datacenter-closed” category. The closed category (as opposed to open) requires submitters to run inference on a given model as-is, without significant software modification. The data center category tests submitters on bulk processing of queries, as opposed to the edge category, where minimizing latency is the focus.

Within each category, there are 9 different benchmarks, for different types of AI tasks. These include popular use cases such as image generation (think Midjourney) and LLM Q&A (think ChatGPT), as well as equally important but less heralded tasks such as image classification, object detection, and recommendation engines.

This round of the competition included a new benchmark, called Mixture of Experts. This is a growing trend in LLM deployment, where a language model is broken up into several smaller, independent language models, each fine-tuned for a particular task, such as regular conversation, solving math problems, and assisting with coding. The model can direct each query to an appropriate subset of the smaller models, or “experts”. This approach allows for less resource use per query, enabling lower cost and higher throughput.

Within the popular datacenter-closed benchmark, submissions based on Nvidia’s H200 GPUs and GH200 superchips, which combine GPUs and CPUs in the same package still came up as the winner.

On a per accelerator basis, Nvidia’s Blackwell outperforms all previous chip iterations by 2.5x on the LLM Q&A task, the only benchmark it was submitted to. Untether AI’s speedAI240 Preview chip performed almost on-par with H200’s in its only submission task, image recognition. Google’s Trillium performed just over half as well as the H100 and H200s on image generation, and AMD’s Instinct performed about on-par with H100s on the LLM Q&A task.

Why Blackwell Need Attention For ML

If we are into ML especially LLM (Large Language Model), Nvidia Blackwell ability to run the LLM using 4-bit floating-point precision is a differentiator.

Nvidia and its rivals have been driving down the number of bits used to represent data in portions of transformer models like ChatGPT to speed computation. Nvidia introduced 8-bit math with the H100, and this submission marks the first demonstration of 4-bit math on MLPerf benchmarks.

By using low-precision numbers without sacrificing accuracy would be a challenge, so Blackwell’s success is it is almost doubled memory bandwidth, 8 terabytes/second, compared to H200’s 4.8 terabytes/second.

Technical Analysis - Multi-timeframe and MACD

Nvidia is still showing strong upward trend from MACD, and MTF is also giving Strong upside, Nvidia is trading above the short-term and long-term MA.

I am expecting a surge when we are near the first shipment of the Blackwell and implement into a data center.

Technical Analysis - SuperTrend for Price Target

If we looked the chart from Supertrend and there is chance that Nvidia might go down to $107 (the low price target but if we observed the green line formed, if we saw an upward movement then we could be seeing Nvidia moving towards the $160 price level.

So I would be watching Nvidia closely and see if it is time to keep loading Nvidia, as I currently hold some shares.

Summary

I am holding onto Nvidia for a better upside, as Blackwell would be a differentiator for Nvidia in the AI inference competition.

There might be more news coming as all the chip makers continue to better their chips and software for the AI inference.

Appreciate if you could share your thoughts in the comment section whether you think Nvidia would extend its leadership in AI chips especially on the AI inference area?

@TigerStars @Daily_Discussion @Tiger_Earnings @TigerWire appreciate if you could feature this article so that fellow tiger would benefit from my investing and trading thoughts.

Disclaimer: The analysis and result presented does not recommend or suggest any investing in the said stock. This is purely for Analysis.