How many loans to negative cash flow start ups were made using Nvidia chips as collateral?
Source: Edward Dowd @DowdEdward
Here are the key takeaways from a great post by Kris Patel on X: 🔥 $GOOGL vs. $NVDA — The Market is mispricing the AI War.
Everyone’s arguing about who has the fastest chip. But that’s not the real disruption. Google isn’t trying to beat Nvidia on speed. Google is trying to beat Nvidia on economics. Here’s the overlooked truth: 1️⃣ The “Nvidia Tax” Nvidia sells to hyperscalers with 70%+ margins. AWS, Azure, and everyone else pay it, and pass it on to you. 2️⃣ Google plays a different game They’re the only hyperscaler that doesn’t need to profit from selling chips. They build TPUs at cost and control the entire stack: Chip → Interconnect → Data Center → Cloud. No margin stacking. No tax. 3️⃣ Training ≠ Inference Training = Ferrari (Nvidia dominates). Inference = Semi-truck (cheap, reliable, scaled). As AI matures, ~90% of spend shifts to inference. And here’s the bombshell: 👉 If Google drives cost-per-token toward zero with TPUs + aggressive cloud pricing… Raw speed stops mattering. Economics wins. Nvidia is selling generators. Google is building the electric grid. Cheap compute + massive distribution = 🏰 Empire. Source: Kris Patel
Nvidia responds to news of Meta using Google's TPUs, sending $NVDA stock -6% lower:
Nvidia says they are "delighted by Google's success" and they "continue to supply Google." They also say, "Nvidia is a generation ahead of the industry" and "offers greater performance, versatility, and fungibility." The AI wars are heating up. Source: The Kobeissi Letter
Nvidia: What’s the Bull Case? What’s the Bear Case?
🐂 Bull Case: The AI infrastructure boom is still in its early innings. Demand has no visible ceiling, and Nvidia continues to command extraordinary pricing power despite new entrants. Every major hyperscaler, enterprise, and startup is still rushing to deploy more compute. 🐻 Bear Case: Gross margins have slipped from 78% → 73%. The product transition introduces execution risk. And if hyperscalers don’t see the ROI they expect, they could slow down spending — even temporarily. What does Nvidia say? On the earnings call, the CFO addressed it directly: ➡️ AI demand remains extremely strong — strong enough that Nvidia is raising Q4 guidance again. Source: Bloomberg Activate to view larger image,
We are seeing a lot of threads on X talking about Nvidia using financial engineering to boost results.
See one example below from The Coastal Journal.
24 years ago, Nvidia replaced Enron in the S&P 500...
Similar to last year anniversary, it seems that some investors see this as an ominous sign...
Nvidia $NVDA said sales to just its number one largest customer represented 22% of its total $57 Billion of revenue during Q3
That means someone spent $12.5 Billion with Nvidia during the quarter 🤯 61% of Nvidia $NVDA Q3 revenue came from just 4 customers: $12.54B – Customer A (22%) $8.55B – Customer B (15%) $7.41B – Customer C (13%) $6.27B – Customer D (11%) So who are A, B, C, D in your view ??? Source: Wall St Engine on X, Evan on X
Nvidia $NVDA saves humanity again with blowout earnings... 🚀
Q3 REV: $57.01B vs $55.19B exp. Q3 EPS: $1.30 vs $1.25 exp Q4 REV: $63.7B TO $66.3B vs $62B exp “Blackwell sales are off the charts, and cloud GPUs are sold out.” — Jensen Huang, Q3 FY26 earnings release (Nov 19, 2025) $NVDA is up +6% after-hours... The market is saved
