Executive Summary
Market Consensus: Nvidia owns AI infrastructure with insurmountable moat. Variant Perception: Google's seventh-generation TPUs have achieved performance parity at dramatically lower cost, creating the first viable alternative for hyperscale deployments. Meta's willingness to partner with Google despite fierce advertising competition validates this technical breakthrough. While Nvidia commands $30-35B per gigawatt, Google's integrated approach offers superior efficiency. The $600B+ CapEx cycle through 2028 creates massive TAM for diversification. Google trades at 1.7x PEG vs historical premium, suggesting market hasn't recognized TPU monetization potential. Recent insider selling reflects routine diversification, not fundamental concerns, given consistent earnings beats. This represents a structural shift from Nvidia monopoly to duopoly, with Google capturing meaningful share of the largest infrastructure build-out in history.
Key Insights
what Mandeep Singh said“Google is in the seventh generation of their TPUs. So clearly they have been working at this for a while, which is the reason why the TPU has been able to catch up to Nvidia's GPU in terms of performance.”
what Mandeep Singh said“Mark Zuckerberg has mentioned that they plan to spend up to $600 billion in CapEx through 2028. So when you are spending such big sums, you don't want to be dependent on one supplier.”
what Mandeep Singh said“They don't want to be giving Nvidia, you know, 300 billion dollars just for the chips. They want to diversify and really make sure they're running their infrastructure at the lowest cost, which is what Google is doing.”
This is a preview. Log in to see the full analysis including investment opportunities, risks, catalysts, and detailed insights.