Nvidia JUST Proved the AI Stock Bubble is about to Burst.
Overall Thesis
Nvidia's AI growth guidance is artificially inflated by circular capital flows from Google and SpaceX fundraising, suggesting the AI bubble is expanding rather than justified by fundamentals.
Narratives
Kevin notes Google raised $32 billion in a single day in February 2026 and $80 billion by June for AI infrastructure, including a rare 100-year UK bond offering. He frames this capital raising as part of a broader 'sucking' of money that ultimately flows into Nvidia hardware spend rather than offering a direct view on Google's stock.
Key Arguments
- Google raised $32 billion in 24 hours in February 2026 for AI infrastructure, including a 100-year bond offering
- By June, Google had raised a cumulative $80 billion
- Google deserves credit alongside OpenAI, Anthropic, and Elon Musk's companies for driving capital into Nvidia hardware spend
Kevin argues Nvidia's massive backlog increase (70% growth guidance vs. 48% prior) coincides suspiciously with over $160 billion raised by Google and SpaceX, suggesting circular financing is inflating AI hardware demand. He remains fundamentally bullish on Nvidia's near-term hardware rally but flags red flags in the underlying economics, including unsustainable Frontier Lab margins and Nvidia guaranteeing deals that recycle back into its own revenue.
Key Arguments
- Google and SpaceX raised over $160 billion in capital which coincidentally matches Nvidia's $160 billion backlog increase
- Nvidia's $105 billion guarantee to OpenAI translates into roughly $82.5 billion of GPU revenue, nearly a full quarter of Nvidia's data center revenue from one customer
- Anthropic's Nscale contract implies compute pricing of about $17.4 billion per gigawatt, below Nebius's $20-25 billion per gigawatt long-term contracts, undermining Elon Musk's $40-50 billion per gigawatt revenue assumptions
- Frontier Labs like Anthropic reportedly add back training costs to inflate margins, which Kevin calls unsustainable economics
Kevin discusses Elon Musk's justification for AI compute revenue projections of $40-50 billion per gigawatt, contrasting this with Anthropic's recently signed $17 billion per gigawatt contract, suggesting Musk's assumptions may be too optimistic. This commentary centers on compute economics tied to Musk's ventures rather than a direct thesis on Tesla stock itself.
Key Arguments
- Elon Musk reportedly justifies chip spending using revenue assumptions of $40-50 billion per gigawatt
- A recent Anthropic/Nscale contract implies pricing closer to $17 billion per gigawatt, well below Musk's assumed figures
- Kevin suggests this mismatch could push break-even timelines for compute investments from under a year to three-to-four years
Hedges & Caveats
- Host acknowledges Nvidia's higher valuation may be justified
- Presenter claims to identify 'red flags' in earnings but transcript cuts off before full analysis
- Host states he is 'not trying to be biased in one direction or another'
- The $160 billion backlog increase coincidentally matches Google/SpaceX capital raises, suggesting correlation rather than causation analysis