So last week was pretty wild if you were paying attention to the AI space. NVIDIA's earnings call basically confirmed what everyone wanted to hear - the Grace Blackwell ramp is happening faster than expected. But here's what stuck with me more: Jensen spent most of his opening remarks drilling into the China export control situation. That's the kind of macro risk that doesn't get enough airtime.

The other thing that caught my eye was Anthropic's Dario Amodei saying barely anyone's talking about the job disruption coming for white collar workers. Entry-level knowledge work is about to get weird. Meanwhile, I noticed something about how these companies training massive language models are basically running out of training data. Google dropping $60M for exclusive Reddit access makes way more sense now - millions of real conversations and product discussions. That's gold for LLM training.

Here's what's actually fascinating though: the real bottleneck isn't just data anymore, it's compute and energy. NVIDIA's building out Omniverse and Cosmos specifically so companies can generate synthetic training data at scale. Billions of iterations in hours instead of months. When you think about it that way, the robotics angle becomes clearer. Figure and the other humanoid companies aren't just building robots - they're building systems that need massive amounts of synthetic world data to train on.

Which brings me to the energy piece. Nobody's really connecting the dots that datacenter buildout is creating insane demand for new power sources. That's why the nuclear play got so hot last week. After Trump's executive orders on reactor permitting and uranium production, you saw LEU jump 18%, SMR spike 25%, and OKLO absolutely soar 29.5%. These stocks are trading at crazy valuations right now - like 100X sales on some of them - but the underlying thesis is solid. The country needs at least 50 new small modular reactors in the next decade just to power the AI infrastructure.

So if you're looking at tech stocks to buy now, you've got multiple angles. The obvious play is still robotics and AI - I've been recommending BOTZ as an easy entry point since NVIDIA's the largest holding and it gives you exposure to Figure and the other serious competitors coming up. But the less obvious play is the nuclear infrastructure piece. NUKZ is worth grabbing as a lower-risk way to track the SMR companies without picking individual names that are this frothy.

The bigger picture is that this AI infrastructure race has three layers: the chips (NVIDIA obviously), the robots that run on those chips, and the power infrastructure to run everything. Most people are only looking at layer one. Tech stocks to buy now need to account for all three. That's why I'm watching both the robotics ETF and the nuclear ETF - they're not sexy individual picks, but they're how you actually position for the next decade of AI buildout.

One thing I missed was how connected some of these dots were. Chris Wright from the Department of Energy was on Oklo's board - that kind of detail matters way more when you're actually tracking the companies. Tech stocks to buy now aren't just about quarterly earnings anymore, they're about infrastructure positioning. If you're serious about AI exposure, you need to think beyond just chip makers and actually consider the full stack. That's where the real opportunity is hiding.