Featured Posts
Markets are settling into holiday mode, with stock futures slightly down on the last full trading day before Christmas. Here’s where things stand today: * Nasdaq-100: +0.37% * S&P 500: +0.04% (flat) * Dow Jones: -0.22% On Friday, the S&P 500 rose 1.1%, recovering
Quantum computing stocks are taking off in 2024. Is this the start of something big? Let’s take a closer look Google’s Chip Changes the Game Google’s new Willow chip, a 105-qubit processor, hit a big milestone: * Fixed Errors: It solved a 30-year problem by cutting down error
Government surveys that track the economy are failing. Budgets are shrinking, people are skipping surveys, and economic data we rely on is becoming less trustworthy. This is a big deal—because when the numbers are wrong, decisions that affect markets, businesses, and everyday life go wrong, too. Here’s what’
The S&P 500 Value Index just dropped 3.7%, marking its longest losing streak on record. Meanwhile, Bitcoin smashed past the $100,000 mark, and the Nasdaq climbed to 20,000. To me, it seems that value stocks have been left in the dust, and investors chasing bigger
Nvidia’s earnings have sparked an AI frenzy, driving record investments. But with soaring valuations, is this boom sustainable or a bubble set to burst?
Remember when we all thought AI was going to change the world last summer? Well, the generative AI wave has only gotten crazier.
Nvidia’s older Q2 earnings guide not only hit the mark but smashed right through it, sending the market into a frenzy over GPUs.
Before Nvidia’s bombshell, AI was already making waves with consumer launches like ChatGPT and image creation platforms like Midjourney.
These brought AI into everyday conversations. Nvidia’s results gave businesses hard proof that AI is a real money-making tool, capable of generating billions in revenue.
Now, investors are piling into AI like it's the next gold rush, driving record valuations. But there's a question we aren’t asking: What exactly are all these GPUs being used for? And more importantly, who is the customer’s customer?
The burning question here is: How much value needs to be generated for this rapid investment to pay off?
To put it simply, take Nvidia’s run-rate revenue forecast and multiply it by 2x to reflect the total cost of AI data centers (GPUs are only half the total cost, the other half includes energy, buildings, etc.). Then, double it again to reflect a 50% gross margin for the end-user of the GPU (like startups or businesses using Azure or AWS).
What’s Changed Recently?
- Supply Shortage Subsided: Late 2023 was peak GPU supply crisis. Startups were desperately calling Venture Capital firms for money to buy GPUs. Now, it’s much easier to get them with quick lead times.
- Growing GPU Stockpiles: Nvidia reported that about half of its data center revenue came from big cloud providers. Microsoft alone likely represented about 22% of Nvidia’s Q4 revenue.
- The $600B Question: Even with generous assumptions for big tech’s AI revenue, there’s a massive gap to fill each year to justify current CapEx levels.
- Nvidia’s B100 Chip: Nvidia’s new B100 chip promises 2.5x better performance for only 25% more cost. This will likely lead to another surge in demand, causing a temporary supply shortage.
One argument you'll hear often is that “GPU CapEx is like building railroads” – the infrastructure is being built, and the value will come. While I agree with this to some extent, there are critical differences:
- Lack of Pricing Power: Unlike railroads, GPU data centers have much less pricing power. New entrants keep flooding the market, and prices are being driven down to marginal cost.
- Depreciation: Semiconductors RAPIDLY depreciate. Nvidia will keep releasing better chips like the B100, devaluing previous generations quickly.
The AI hype is reaching new highs, but the road ahead is long.
Founders focused on delivering real value to end users will capture the most value from this overbuilt infrastructure.
AI has the potential to be a generation-defining technology wave, and companies like Nvidia deserve credit for enabling this transition.
So, let's not get carried away by the delusion that AGI is coming tomorrow or that we need to stockpile GPUs.