To The Moon

Here are my talk about AI Industry. Full live link replay >>

I have shared a lot about AI trends in the past. Just bear in mind that the AI total addressable market is huge.

The valuation for a lot of AI stocks is expensive, but still not as crazily expensive as during the dot-com bubble.

For example, the P/E ratio for the Nasdaq 100 is now trading at 35x. At the peak of the dot-com bubble, it was trading at 175x.

1995 marked the start of internet monetization. It also kickstarted the dot-com bubble.

The entire dot-com bubble lasted for 5 years, from 1995 to 2000.

At its peak, the $.SPX(.SPX)$ returned 265%.

If AI is a bubble similar to the dot-com bubble, maybe it can continue for another 5 years, starting in 2024.

So far, the $.SPX(.SPX)$ has only gone up by 24%.

$NASDAQ 100(NDX)$ returned more than 1000% during the dot com bubble. And $NASDAQ 100(NDX)$ only returned 27% so far in 2024.

This is the AI value chain.

Recently, $Broadcom(AVGO)$ went up 24%, but $NVIDIA(NVDA)$ , on the other hand, went down 6%.

I reckon the reason is that some investors are selling off their $NVIDIA(NVDA)$ holdings and investing in $Broadcom(AVGO)$ .

Investors likely believe that ASICs, Broadcom's custom AI chips, can replace Nvidia's GPUs.

I reckon they are wrong. Custom AI chips can complement Nvidia's GPUs, but they can't replace them.

For example, Have you ever wondered why Amazon's custom AI chips, Trainium, are sold at a 40% discount compared to Nvidia's GPUs?

And why, why didn’t $Amazon.com(AMZN)$ sell them at a higher price?

It is because $NVIDIA(NVDA)$ GPUs are still the best AI chips for AI training. Most companies prefer Nvidia’s AI chips for training because if they don’t, they risk falling behind in AI performance.

All in all, I am constructive about the AI inference market as well. The main beneficiaries include $Broadcom(AVGO)$ , $ARM Holdings(ARM)$ , $Microsoft(MSFT)$ , $Alphabet(GOOG)$ , $Amazon.com(AMZN)$ , and $Taiwan Semiconductor Manufacturing(TSM)$.

This is the AI S-curve in 4 phases:

• Phase 1: InfraThis is the AI S-curve in 4 phases: AI core structure: GPU, Datacenter, Power, Cloud.

• Phase 2: Edge AI, which includes AI smartphones and AI PCs

• Phase 3: AI agents, like co-pilots

• Phase 4: Physical AI, like self-driving cars and AI robots 

Below are the companies that should benefit in Phase 1 and Phase 2.

Phase 1: Infrastructure

•Chips: $NVIDIA(NVDA)$ , $Broadcom(AVGO)$ , $Taiwan Semiconductor Manufacturing(TSM)$ , $Marvell Technology(MRVL)$ , $ARM Holdings(ARM)$ , $Cadence Design(CDNS)$ , $Advanced Micro Devices(AMD)$ , $Synopsys(SNPS)$

•Data Centers: $Oracle(ORCL)$ $Amazon.com(AMZN)$ , $Alphabet(GOOG)$ , $Dell Technologies Inc.(DELL)$ , $Microsoft(MSFT)$

•Power: $Vistra Energy Corp.(VST)$ , $Constellation Energy Corp(CEG)$ , $BWX Technologies Inc(BWXT)$ , $Oklo Inc.(OKLO)$

•Data: $Salesforce.com(CRM)$ , $Palantir Technologies Inc.(PLTR)$ , $Snowflake(SNOW)$

Phase 2: Edge AI

• $Dell Technologies Inc.(DELL)$ , $Taiwan Semiconductor Manufacturing(TSM)$ , $Qualcomm(QCOM)$ , $Apple(AAPL)$

At this moment, market participants seem to be buying all companies across all 4 phases.

Phase 3: Agentic AI

• $Microsoft(MSFT)$ , $NVIDIA(NVDA)$ , $Salesforce.com(CRM)$ , $Meta Platforms, Inc.(META)$ , $Adobe(ADBE)$ , $Shopify(SHOP)$ , $Amazon.com(AMZN)$ , $Alphabet(GOOG)$ , $Palantir Technologies Inc.(PLTR)$

Phase 4: Physical AI

• $Tesla Motors(TSLA)$ , $Alphabet(GOOG)$ , $NVIDIA(NVDA)$ , $Amazon.com(AMZN)$ , $Intuitive Surgical(ISRG)$

But I think you'll hear more about AI agents in 2025.

I believe AI software companies could probably outperform in 2025.

If you're unfamiliar with the names under these 4 phases, just remember that the "Magnificent Seven" stocks should benefit from this AI boom.