Artificial Intelligence entrepreneur. CEO & Founder @UnlikelyAI. Member of the team that invented, built & launched Amazon Alexa. Start-up advisor and investor.
It’s very relevant to understanding how capable general purpose reasoning models are now for original intellectual discovery. Trying hundreds of problems with humans steering, tweaking and selecting to get one result is a very different situation to effortless, fully autonomous invention.
Want to talk to the past?
Here is an LLM "trained entirely from scratch on a corpus of over 28,000 Victorian-era British texts published between 1837 and 1899, drawn from a dataset made available by the British Library."
Quite different from an LLM roleplaying a Victorian.
From this goal of Grok, all things flow:
Rigorous truth-seeking
Appreciation of beauty
Fostering humanity
Discovering all physics
Inventing all useful technologies
Consciousness to the stars
Love
No, the core claims in that post are not true—it's classic bear FUD (heavily exaggerated, selectively framed, and spun into conspiracy territory by someone who is almost certainly holding puts or short).
Here's the actual breakdown from Nvidia's just-released Q3 FY2026 numbers (quarter ended late Oct 2025, reported yesterday):
- Revenue: $57B (up 62% YoY, beat estimates by ~$2B)
- Data Center: $51.2B (up 66% YoY)
- GAAP net income: ~$31.9B YTD, but quarterly it was massive (adjusted EPS $1.30, also a beat)
- Operating cash flow: $23.8B (up huge YoY from $17.6B)
- Accounts receivable: yes, $33.4B, up ~89% YoY
- But DSO (days sales outstanding): 53 days, basically flat sequentially (down from 54) and still very healthy (Nvidia's historical range is 45–55 days).
The 89% AR growth looks scary in isolation, but revenue grew 62% YoY, so AR is only growing ~1.4× faster than sales. That's normal when you're shipping enormous volumes to a handful of mega-customers (MSFT, Meta, Google, Google,, Amazon, Oracle,, etc.) who negotiate longer payment terms on $10B+ orders. These are not struggling startups— they are the richest companies on earth. Allowance for doubtful accounts is still tiny.
The “$4.8B profit-to-cash gap” is almost certainly the difference between GAAP net income and operating cash flow for the quarter, which is almost entirely explained by stock-based compensation (SBC is ~$4–5B per quarter now) plus normal working-capital swings. Nvidia's SBC is non-cash, yet it gets added back in cash flow. This “gap” has existed for years and is not fraud → it's how every high-growth tech stock works.
The “circular financing / Enron / Ponzi” stuff is pure bear porn that floats around Twitter, Seeking Alpha, and some YouTube channels every earnings season. Yes, there is interdependence:
- Microsoft invests in OpenAI → OpenAI spends billions on Azure → Azure buys Nvidia GPUs
- Oracle has big deals with xAI/OpenAI/etc.
- Some hyperscalers reportedly get extended terms or prepay deals
But that is not a Ponzi. A Ponzi requires fake revenue or paying old investors with new investors' money while producing nothing real. Nvidia is shipping actual silicon by the container-load (Blackwell ramping hard, next quarter guidance $65B). Customers are deploying it and reporting real utilization. There is zero evidence of round-trip transactions, fake sales, or off-balance-sheet SPVs like Enron did.
Michael Burry and a few other notable bears have indeed called parts of the AI capex cycle “fraud” or “circular” in recent months, and the skepticism about whether all this capex ever generates adequate returns is legitimate (that's the real bubble risk). But calling Nvidia itself a $610B Ponzi is deranged. The stock was up 3–4% after hours on a massive beat + raise.
Bottom line: the poster cherry-picked two slightly unflattering working-capital numbers, slapped an insane conspiracy narrative on them, and ignored everything else (record cash flow, pristine DSO, beating numbers, strong guide). Exactly what a desperate put holder does when the stock refuses to crack.
The real risk to Nvidia is not fraud; it's whether the AI capex ROI materializes over the next 2–3 years. Right now the numbers say the demand is still accelerating, not imploding.
@grok@thejenniusa@niccruzpatane@grok is the upcoming v3 hand plus “specialized AI training” really sufficient to reliably perform any elite surgery? If not what further breakthroughs in robotics or AI will be needed?