Video summary
Forget Nvidia. This Is the New King of AI
Main summary
Key takeaways
Market / Investing Context (AI Infrastructure)
- The video frames the day as positive for AI investors, citing strong market reaction to recent quarterly earnings and forward guidance from AI “neoclouds” / infrastructure providers.
- It argues Nvidia’s AI dominance is not “over”, despite a stretch of stagnation.
Stocks / Tickers / Assets Mentioned
- NVIDIA (NVDA)
- CoreWeave (appears as “Core Reef” / “Kore” in subtitles)
- Nebius (appears as “Nebuse” / “Nebus”)
- Counterparties mentioned for Nvidia financing / partnerships:
- Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, KKR
- Wall Street Journal referenced for Anthropic IPO timing.
- Other AI labs referenced regarding IPO expectations:
- OpenAI, Anthropic
- SpaceX AI referenced as an example of aggressive scaling.
- No explicit ETFs, bonds, commodities, or crypto tickers were named.
Key Performance Numbers & Company Fundamentals
Nvidia (NVDA)
Price Performance
- YTD: +18.5%
- Previously described as near “flat” (~2 weeks earlier)
- Earlier drawdown cited: ~19%
Valuation Multiples
- Trailing P/E: 34.3x
- Forward P/E: 22.4x
- Price-to-Free-Cash-Flow: 22.7x
Company Scale
- Valued at over $5 trillion
Growth Expectations (fiscal years)
- Fiscal year: revenue growth expected 82%
- Fiscal 28: 42%
- Fiscal 29: ~22–23%
Quarterly Growth Acceleration Claims
- Expects an upcoming quarter in ~2 weeks with >100% YoY revenue growth
- Mentions expected revenue afterwards: >$91B
- Prior growth comparisons referenced: 80% / 70% / 54% / 47% YoY (as referenced in subtitles)
Earnings Beat History
- Says NVDA has consistently beaten analyst expectations
- Analyst expectation cited: $91.8B
- Claims potential beat to around $92B–$93B
- Recent beats referenced (approx. sequence from captions): 3.2%, then close to 3%, then 3.7%, 1.5%, 1.9%
Capital Markets / Infrastructure Financing Announcement
- Nvidia partners with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, KKR (and others implied) to mobilize $500B+ of third-party capital
- Purpose: create an investable AI compute infrastructure “asset class” to broaden access and enable longer-duration usage
- Stated implication: lower cost of capital → faster AI buildout (though data centers still take time)
Nebius (Nebius)
Stock Performance (on the day described)
- Up: 23%
- Around $238 per share
Recent Quarterly Growth
- Revenue: +454% YoY to $582M
- ARR: up to $3B (~600% YoY)
Profitability / Adjusted EBITDA
- Group adjusted EBITDA margin referenced: 50%
- Adjusted EBITDA: $236M (group total)
Pricing / Monetization
- Mentions an asset “glide capacity model” and short-term premium tiers
- Pricing cited: $40–$50M per megawatt for premium capacity
Deal Economics / Payback
- Compressed deal payback: ~1 year 10 months
- Previously: 2–3 years
Guidance / Capacity Expectations
- Expects to end the year with 5 GW of contracted power (previously >4 GW)
- “Violent capacity” appears to refer to a 2026 milestone (not 2027), with:
- Connected power by year-end: 800 MW to 1 GW
- Possible translation into active power mid-2027
- ARR scenario (implied/inferred by narrator): ~$20–$25B ARR in 2027
Capacity Auctions / Price Discovery
- First capacity auction:
- Priced 15% above any prior “Blackwell high” ever cleared
- Introduces an “explicit live price discovery engine”
- Customer satisfaction and desire for more capacity mentioned
Capacity Contract Details
- First 3–6 month premium capacity contracts mentioned in Q3, going live in Q4
- Enables $40–$50M per MW pricing for immediate workloads
Capital-Light / Partnership Model
- Partner provides: land, facility, power infrastructure, hardware, capital, facility ops
- Nebius provides: full-stack software system architecture, go-to-market and demand
- Benefit: partner capacity becomes high-margin ARR with minimal Nebius capex, accelerating growth
Power Generation / Execution
- Uses Bloom Energy fuel cells
- Described as quiet, reliable, ultra-low emission, deployed on site
- Trenching delivered on schedule; building finished earlier in the summer
- Claims confidence that execution stays on time
CoreWeave (CoreWeave)
Stock Performance (on the day described)
- Up: 17.7%
Recent Quarterly Growth
- Revenue: +112% YoY to $2.57B
- +24% QoQ
Backlog / Pipeline
- Revenue backlog: $104.2B
- Says they signed $25B early in Q3
Profitability / EBITDA
- Adjusted EBITDA: $1.51B
- Margin: 59%
- Mentions a prior “gap net loss” of -$626M
Balance Sheet / Interest Rate Drag
- Raised $18B in Q2
- Total capital: over $32B
- Weighted average debt cost reduced by ~300 bps YoY
- Annualized interest savings: ~$1.1B
- Q2 interest expense cited: $640M more than doubled YoY
- Q3 guidance for interest expense: $860M to $940M
Power Scaling
- Over 1.5 GW of active power
- Added ~500 MW active power in Q2 (largest quarterly expansion)
- 51 data centers
- 1 GW contracted outside the US, including:
- 360 MW in APAC (Indonesia) expected online in ~18 months
Non-GPU Services Growth
- “Manage inference”:
- from ~$1M to > $1B ARR in a single quarter
- guiding for > $250M ARR by end of year
- “NonGPU software storage services”:
- > $400M ARR
Pricing Actions
- 25% list price increase across SKUs in July
- Q2 contracts have 5–10 points higher contribution margins, especially for Vera Rubin clusters
GPU Lifecycle Monetization
- Signed a contract for 2020 A100 GPUs extending to 2029
- Framed as retaining and monetizing value from legacy GPU fleets
Capex / Spending and Guidance (as discussed)
- Full-year revenue expected: $12.4B to $13.2B
- Exit run rate: $18.5B to $19.5B
- Capex:
- Nebius: $20B–$25B (reaffirmed)
- CoreWeave: $35B–$39B (narrator says “this fiscal year”)
- Narrator expectation (not formal guidance): Nebius capex could double in 2027
Nvidia Catalyst → “AI Compute Financing Platforms”
The video’s thesis for new infrastructure financing platforms:
- Mobilize $500B+ of third-party capital
- Transform Nvidia compute + full-stack AI infrastructure into an investable asset class
- Enable longer duration usage
- Reduce cost of capital, accelerating AI compute deployment
Upcoming Events / IPO Implications
Anthropic IPO
- Source: Wall Street Journal
- Timing suggested as October
- Current run rate cited: $70B ARR
- Prior raise referenced as $965B (captioning error possible; narrator implies it may be $965M)
- Target IPO size: $1.5T (framed as “SpaceX number,” implying extremely large valuation)
Expected Market Pattern
- Run-up into IPO, followed by a valuation crash
- Compared to SpaceX and “space stocks”
- Potential volatility driven by investors demanding more detail on:
- profitability
- growth
- burn
- model-level economics
- model examples cited: “Fable 5, Opus 4.8, Sonnet 5” (likely Anthropic)
Explicit Recommendations / Cautions / Stance
- No formal “buy/sell” order is issued.
- The narrator’s stance is broadly bullish:
- Nebius and CoreWeave quarters are described as excellent
- suggests momentum in growth and profitability
- Implicit caution:
- Volatility expected around Anthropic/OpenAI IPOs
- narrator opinion: if public earlier (e.g., July), they “would have been cut by 50%”
Disclosures / Disclaimer
- No clear “not financial advice” disclaimer appears in the subtitles provided.
Methodology / Framework Mentioned
- No explicit step-by-step portfolio construction framework.
- The video implicitly uses an “AI infrastructure value chain / monetization” framework:
- Chip/platform dominance (NVDA + CUDA lock-in)
- Financing + lower cost of capital
- Capacity scaling measured in MW/GW
- Monetization via ARR, including:
- premium tiers ($40–$50M/MW)
- non-GPU services (inference, storage)
- partnership/asset-light models to reduce capex intensity
- backlog/run-rate as demand indicators
- Profitability metrics: adjusted EBITDA and margins
- Balance sheet risk: interest expense and debt costs (bps), plus annualized interest savings
Presenters / Sources
- Presenter: not explicitly named in the subtitles
- News/source referenced: Wall Street Journal (Anthropic IPO timing)
Rate this summary
Your feedback will help improve summaries.
Improve this summary
Reprocess with a stronger model when the summary feels incomplete or inaccurate.
Translate summary in another language
Ask questions to this video
Chat for follow-up questions, clarifications, and source-backed answers.