Video summary

GET IN EARLY! These 3 Stocks Will Make Millionaires By 2029

Main summary

Key takeaways

Finance

Finance-Focused Summary (AI Infrastructure / “Neoclouds”)

The presenter argues that Nvidia’s latest earnings indicate AI infrastructure demand is accelerating and broadening beyond the largest hyperscalers. He proposes three “Neocloud” stocks—CoreWeave, Nebius, and Iron (Iron/Bitfarms’ former mining entity positioned as an AI cloud)—as major beneficiaries through 2029, with capacity expansion targets into 2026–2027.

Macro / Market Framing

The presenter interprets Nvidia’s growth/earnings as evidence that:

  • AI spending is not slowing
  • AI infrastructure buildout is accelerating
  • Spending is expanding beyond “trillion-dollar tech giants” (hyperscalers) into a wider set of customers (the ACIE category)

Method highlighted

To monitor AI data center operators, the presenter emphasizes:

  • Revenue growth
  • Especially: capacity expansion
  • Additional checks: backlog
  • And funding/interest costs (interest burden)

Key Ticketers / Companies / Instruments Mentioned

Public equities (tickers)

  • Nvidia (NVDA)
  • Alphabet (Google) (no ticker given)
  • AMD (no ticker given)
  • CoreWeave (CRWV)
  • Nebius (NBIS)
  • Iron (AI cloud entity: IRN)

“Tickerol U” is mentioned as a channel/host name, not a financial instrument.

Other companies referenced (no tickers given)

  • Microsoft, Amazon, Meta Platforms, Google, OpenAI, Perplexity, Figure AI
  • Alphabet, AMD also appear as profitability/comparison benchmarks

Notable Technology / Infrastructure Terms

  • Nvidia platform / infrastructure:
    • Blackwell racks
    • Bluefield 4 GPUs
    • Next-gen networking
  • Nvidia compute:
    • Vera Rubin
  • Nvidia inference chip mentioned:
    • Gro 3 LPX inference chips
  • Power/capacity units and concepts:
    • GWatt (gigawatts), MW (megawatts)
    • Active vs contracted power

Nvidia Earnings Metrics and Implied AI Buildout Path

  • Earnings date referenced: August 26
  • Quarterly revenue: $96.2B
    • +18% QoQ
    • +106% YoY
  • Operating income: +124%
  • Operating margin: 66.2%
    • Comparisons cited: Alphabet 34%, AMD ~17%
  • Operating expense growth: +55% (about half of revenue growth)
  • EPS: more than doubled
    • Presenter claims ~15% of EPS growth came from $7.8B investment gains (vs $2.2B prior year)

Supply capacity commitments

  • Increased from $119B (last quarter) to $279B (today)
  • Purpose: lock memory for:
    • Vera Rubin (this year)
    • Rubin Ultra (2027)

Reporting Change Used as the Thesis: Hyperscale vs “ACIE”

The presenter says Nvidia changed how it reports data center revenues:

  • Previously: one consolidated number
  • Now split into:
    • Hyperscalers
    • ACIE = AI Clouds, Industrial and Enterprise

Growth rates mentioned

  • ACIE: +138% YoY
  • Hyperscalers: +102% YoY
  • Claim: this is the first time hyperscalers weren’t the fastest-growing segment—signaling broader AI buildout beyond big tech.

Investment selection implication

The presenter focuses on Neoclouds within ACIE—naming:

  • CoreWeave
  • Nebius
  • Iron

Framework / Step-by-Step Approach (As Presented)

  • Use Nvidia’s segment shift (Hyperscale vs ACIE) to find where spend is accelerating (ACIE).
  • For Neoclouds, prioritize:
    • Capacity growth
      • active power and contracted power
    • Revenue backlog
    • Funding structure / interest burden
      • Net debt = total debt − cash
      • Interest expense as % of revenue
    • Customer quality & prepayments
      • look for large hyperscaler/enterprise contracts
  • For valuation comparisons:
    • Prefer enterprise value (EV) over market cap for debt-heavy operators
    • Use EV vs run-rate revenue
    • Use cost per MW (active and contracted) as an execution proxy

CoreWeave (CRWV): “Biggest,” Backlog-Driven

Scale / Capacity

  • Data centers: 51 (North America + Europe)
  • Active power: 1.5 GW
  • Contracted power: 4.2 GW

Active power generates revenue; contracted power represents future electricity availability.

Nvidia compute mapping (presenter’s claims)

  • Nvidia Blackwell racks: 72 GPUs per rack
  • Power per rack: ~120 kW
  • Claimed implication: CoreWeave contracted capacity could power 30,000+ racks or 2.2M+ Blackwell GPUs

Revenue + Backlog

  • Quarterly revenue: $2.6B (+112% YoY)
  • Revenue backlog: $14B (+246% YoY)
  • Backlog drivers cited:
    • multi-year deals involving OpenAI, Nvidia, Microsoft, Meta
    • additional citation: Meta signed $21B with CoreWeave (March)

Nvidia partnership detail (as referenced)

  • Nvidia purchases unused capacity:
    • $6.3B partnership
    • Nvidia buys unsold cloud capacity through April 2032
  • Nvidia investment cited:
    • $2B into Core stock at $87/share
    • presenter states this is higher than current trading price (current price not specified)

Debt / Interest Risk Metrics

  • Debt: $35B
  • Lease obligations: $16B
  • Cash: $5.5B
  • Approx. net debt: ~$46B
  • Interest expense last quarter: $640M (~25% of revenue)

Losses (presenter framing):

  • Operating loss: $49M
  • Net loss: $626M
  • Claim: difference is “almost entirely interest.”

Presenter’s implied take

  • Nvidia’s role strengthens CoreWeave (supplier + launch platform + buyer of last resort), but:
    • Core cannot control interest expense
    • leverage is the key risk

Nebius (NBIS): Strongest Balance Sheet + Fastest Growth (More Expensive)

Product / Platform Positioning

  • Described as “most technically advanced”
  • Workloads:
    • Aether = GPU rental / training cloud
    • Token Factory = inference on top

Nvidia Chip Timing

  • Presenter cites Nvidia CFO:
    • Nebius expected to be first to receive Gro 3 LPX inference chips in volume

Capacity Targets

  • Contracted power: raised 4 GW → 5 GW by end of 2026
  • Active power: expected 800 MW to 1 GW by end of the year
  • Claim: they expect to sell every watt

Booked Contracted Revenue / Prepayments

  • Contracted work backlog: $37.5B
    • includes $17.44B deal with Microsoft
    • includes contract with Meta worth up to $27B
  • Funding model advantage:
    • ~70% of deals in the quarter include customer prepayments
    • prepayments cover 50–60% of equipment costs
    • expected >$9B prepayments this year

Revenue Growth + Guidance

  • Revenue: $582M (+454% YoY)
  • Guidance: $7B–$9B annualized run rate by end of 2026
  • Presenter characterizes as ~7x from last year

Debt / Interest Metrics

  • Debt: $8.5B
  • Lease obligations: $1.5B
  • Cash: $8B
  • Approx. net debt: ~$2B
  • Interest last quarter: $119M (~20% of revenue)

Funding tradeoff noted:

  • possible stock sales → dilution risk

Iron (IRN): Pivot From Bitcoin Mining to AI Cloud

Revenue Transition / Growth

  • Quarterly AI cloud revenue: $70.5M
    • exceeds Bitcoin mining revenue for the first time
  • Prior quarter AI: $33.6M
    • so more than doubled in ~90 days
  • Total revenue slightly down: $145M → $137M
    • attributed to pivot accounting/turnover of Bitcoin hardware

Contracts / Pipeline

  • Pipeline: >5 GW lined up for AI data centers
  • Contracts cited:
    • Microsoft: 5-year $9.7B
    • Nvidia: $3.44B deal

Balance Sheet Metrics

  • Debt: $7.8B
  • Cash: $5.9B
  • Net debt: ~$1.9B
  • Interest last quarter: $25M (~18% of revenues)
  • Backlog: $16.6B
  • Target annualized run rate from capacity this year: $4B
    • presenter says ~$1B already online
    • expects quadruple within next 4 months
    • caution: management suggests recognized revenue could be lower

Losses / Impairment Risk

  • Huge quarterly loss: $684M
    • mostly non-cash impairment from writing off Bitcoin hardware during the conversion to AI

Presenter’s implied take

  • Execution risk centers on:
    • whether contracted power becomes recognized AI revenue fast enough
    • whether conversion costs are too high

Cross-Company Comparison (Valuation + Execution Risk)

The presenter builds a comparison table and stresses it is not audited, due to:

  • different fiscal calendars
  • different contract lengths
  • scaling from different starting points

Enterprise Value vs Market Cap

  • CoreWeave debt is large relative to market cap → use EV
  • Claim: CoreWeave EV ends up ~2x bigger because of debt

“Cheapest” by Power Cost (Execution Proxy)

  • Cost per active MW:
    • CoreWeave: ~$62M per MW
    • Nebius & Iron: > $350M per MW
    • conclusion: CoreWeave ~6x cheaper by active power
  • Cost per contracted MW:
    • Iron: ~$3M per MW (cheapest)
    • Nebius: ~$12M per MW
    • CoreWeave: ~$22M per MW
    • conclusion: Iron ~7x cheaper than CoreWeave by contracted power

Valuation vs Run-Rate (as stated)

  • EV/run-rate multiple (year-end run rate):
    • Nebius: 7.4x
    • CoreWeave: 4.9x
    • Iron: 4x
  • EV sizes cited:
    • Iron: $16B enterprise value (smallest/cheapest)
  • Run-rate scale cited:
    • Iron AI run rate: ~$280M (excl. Bitcoin)
    • Total revenue run-rate: ~$500M

Explicit Recommendations / Selection Guidance (As Presented)

  • If you want the lowest execution risk / want active power already running:

    • Pick CoreWeave (CRWV)
    • Rationale: biggest active power, deepest backlog, and Nvidia “buyer of last resort” through April 2032

  • If you want maximum upside:

    • Pick Iron (IRN)
    • Rationale: cheaper contracted-power economics and relatively low interest burden
    • Biggest risk: converting contracted power into recognized revenue quickly enough without pivot costs overwhelming the economics

  • “Middle” option: Nebius (NBIS)

    • strong growth + better balance sheet
    • presenter notes: more expensive and potential dilution risk

Disclosures / Disclaimers

  • No explicit “not financial advice” disclaimer appears in the provided subtitles excerpt.
  • A sponsor for a voice-to-text tool is mentioned (non-finance), and the presenter’s finance thesis excerpt does not include a direct legal disclaimer.

Presenter / Sources

  • Presenter: Alex (also signs as “Alex,” “Tickerol U”)
  • Channel name used at the end: Tickerol U

Original video