The Nvidia & Jensen Market Collapse Begins.
Meet Kevin
4,011 views • 23 hours ago Save 20 min 4 min read
Video Summary
A prominent AI research firm, Semi-Analysis, has triggered market panic by slashing its 2030 AI capital expenditure forecast by $1.6 trillion, just 11 days after reiterating its previously bullish stance. This sudden about-face, which includes a 66% reduction in projected NVIDIA data center revenue, appears to be a reactionary attempt to align with falling stock prices rather than a reflection of shifts in physical hardware demand or order backlogs.
While market observers grapple with the implications of the Jensen Huang leadership crisis and potential data center delays, the broader AI infrastructure demand remains robust. Despite the aggressive downward revisions and conspiracy-laden speculation surrounding NVIDIA's market position, major financial institutions like Morgan Stanley continue to view the AI spending cycle as intact, suggesting the recent market volatility may be more driven by sentiment and fear than by fundamental industry collapse.
Short Highlights
- The $1.6 trillion reduction in AI spending forecasts by Semi-Analysis is widely viewed as a reactionary move to justify current market sell-offs rather than a reflection of actual hardware demand.
- Physical hardware orders and backlogs across the industry remain at or above pre-revision levels, contradicting the narrative of an immediate collapse in spending.
- The revision relies on questionable assumptions, including a 50% cut in power consumption estimates and a 66% drop in NVIDIA's 2030 data center revenue.
- Major institutions, including Morgan Stanley, maintain that the AI capital expenditure cycle remains intact despite short-term economic headwinds.
- The market is currently absorbing multiple stressors, including the Jensen Huang leadership scandal, Federal Reserve hawkishness, and election-related data center delays.
- Upcoming critical events include NVIDIA's earnings report on November 17th and the Anthropic IPO on November 21st, both of which are expected to dictate market direction.
Key Details
The $1.6 Trillion Spending Cut [0:35]
- Semi-Analysis drastically reduced its AI capital expenditure forecasts for 2027 through 2030, cutting a total of $1.6 trillion from previous projections.
- The firm reduced its 2030 forecast by 40%, moving from a previously estimated $1.53 trillion to $923 billion.
In other words, an extra $1.6 trillion they had forecast, gone, zero, baby. This is insane.
Questionable Model Integrity [4:40]
- The firm reiterated its bullish stance just 11 days prior to the massive downward revision, raising questions about the reliability of their models.
- There is no evidence of a change in physical hardware orders or backlogs in the last 11 days to justify such a significant shift in outlook.
If you've got to adjust your models from one and a half trillion or, you know, you've got to cut $1.6 trillion out of your models in 11 days, your models must not be that good.
The Power and Margin Assumptions [7:15]
- The revised analysis assumes a 50% reduction in power consumption per socket and significantly lower memory prices, which contradicts other market forecasts.
- The model predicts that NVIDIA’s future margins will collapse as companies shift toward cheaper, in-house ASIC alternatives.
They literally walk back their fantasy per-chip power numbers from 12,200 watts to 16,000 watts per socket that all of a sudden can't be plugged in anywhere.
Market Stressors and Sentiment [9:20]
- The stock market is currently navigating a "perfect storm" of negative factors, including the failure to secure a deal with Iran and a hawkish shift from the Federal Reserve.
- Sentiment has been further soured by the Jensen Huang leadership controversy and rising credit default swaps for major cloud providers.
You've got a lot of pain that this market now has to absorb.
The Kitchen Analogy for Silicon [11:10]
- The speaker uses a kitchen analogy to explain the shift, noting that while physical "kitchen size" (silicon area) is only being reduced by 15%, the model assumes a 50% drop in power consumption.
- This discrepancy suggests the research firm is forcing data to fit a narrative of decline rather than reflecting physical constraints.
The bigger your pantry is, the bigger your refrigerator door, the more bandwidth you have to get stuff done.
Institutional Contradictions [15:45]
- While Semi-Analysis is bearish, Morgan Stanley maintains that the AI CapEx cycle is resilient and supported by a solid global economy.
- Other firms like Bernstein argue that high-bandwidth memory prices will continue to grow, directly opposing the assumptions made in the recent downward revision.
The AI CapEx cycle we view as intact, a resilient global economy and solid earnings growth extending into 2027 provide a constructive foundation.
Upcoming Market Catalysts [14:40]
- Key dates ahead include NVIDIA's earnings report on November 17th and the Anthropic IPO on November 21st, which will serve as major tests for market confidence.
- The market will also look toward the December 8th-9th Federal Reserve meeting for further guidance on interest rate policy.
If Anthropic tanks, the whole market's going to keep tanking.