Been seeing a lot of noise around trillion-dollar AI plays lately, and honestly Jim Cramer's recent calls on Alphabet and Amazon are worth paying attention to. Not because Jim Cramer is always right, but because the fundamentals he's highlighting actually check out when you dig into the numbers.

Let me break down what's interesting here. Alphabet's cloud business has accelerated revenue growth for three straight quarters now. That's not nothing. But here's what most people miss - their custom AI chips, the TPUs, are starting to become a real revenue driver. They're not just using these internally anymore. Meta and Anthropic are already renting capacity, and they've got deals with major investment firms for TPU-based cloud services. If that business scales even half as well as people think, it changes the valuation story completely.

Then there's Amazon. AWS still dominates with 41% of the cloud market, and cloud revenue just hit a 13-quarter high at 24% growth. But the real leverage play is how they're using AI to squeeze margins in retail. Generative AI optimizing inventory, delivery routes, the whole supply chain - that's where the compounding happens. And they're also building custom chips called Trainium and Inferentia. OpenAI just committed to consuming 2 gigawatts of Trainium capacity in a deal worth about $138 billion. CEO Andy Jassy says their custom chip business is already running at $10 billion annually and growing triple digits.

Now, Jim Cramer made his calls when Alphabet was around $344 and Amazon around $239. Both have pulled back since then, which is exactly when people get nervous. But here's the thing - Wall Street analysts have Alphabet at a $385 median price target (29% upside from current levels) and Amazon at $285 (31% upside). And both companies have a track record of beating estimates by massive margins - Alphabet by 15% on average, Amazon by 19%.

The valuation concern is real though. Alphabet trades at 28x earnings, Amazon at 30x. That's not cheap. But if you're betting on AI infrastructure becoming as critical as people think, and these two companies are positioned to capture disproportionate value, then current prices might actually look cheap in a few years. Especially when Jim Cramer's thesis is backed up by actual revenue acceleration and custom chip monetization.

Worth thinking about if you're building an AI-focused portfolio.