Forward P/E of 17x is 'valuation hope' not 'valuation discipline'

• Forward EPS: $12.89 vs TTM EPS: $6.53
• EPS expansion: 97%
• TTM P/E: 33.3x
• Forward P/E: 16.9x
• PEG: 0.6
• If consensus cuts 15%, forward EPS falls to ~$10.95
• Forward P/E becomes ~19.9x at current price
• PEG becomes ~1.3
• Market cap: $5.27 trillion
• Beta: 2.22
• The TTM P/E is 33.3x—that's the multiple the market is actually paying today, on actual earnings

↩ rebuts: “Bull's claim that 16.9x forward P/E with PEG of 0.6 is 'essentially a mispricing artifact'”

Inventory is growing 40+ percentage points faster than revenue—a demand-pull signature at best, or channel-stuffing at worst

• Q1 FY26 (April 2025): inventory $11.33B, revenue $44.06B, ratio 25.7%
• Q1 FY27 (April 2026): inventory $25.80B, revenue $81.62B, ratio 31.6%
• Revenue grew: +85% YoY
• Inventory grew: +127% YoY
• Receivables grew: +84% YoY
• Bull claims 'inventory is 8% of revenue'—but only because revenue is being projected forward, not measured
• Inventory days outstanding is the right metric, and on that basis inventory is lengthening, not stable

↩ rebuts: “Bull's 'inventory is 8% of revenue, within historical norms' argument”

$500B third-party capital mobilization signals customer capex limits have been hit—not TAM expansion

• $500 billion of external capital being mobilized to fund AI buildout
• If customers needed outside capital to fund NVIDIA purchases three years ago, it would have been front-page news
• Microsoft, Meta, Google, Amazon—trillion-dollar-balance-sheet companies—if they needed outside capital to fund NVIDIA purchases, it would have been news
• 86% probability of no Fed cuts in 2026 (Polymarket)
• Sovereign buyers are lumpy, politically complex, and often contingent on local manufacturing or equity participation requirements
• Sovereign capex is first line item to get deferred when energy revenues get squeezed (Hormuz risk)

↩ rebuts: “Bull's TAM expansion framing of third-party capital”

Intel raising $20B, AMD MI series gaining traction, and custom silicon from four hyperscalers pose material revenue and operating income risk

• Intel raising $20 billion specifically to re-enter AI accelerators
• If Intel captures 10% of NVIDIA's data center share over next 3 years: ~$25-30B revenue at risk
• At 65% operating margin: $16-20B operating income at risk
• 30% of TTM operating income at risk
• AMD MI300X/MI325X deployed at scale by Microsoft, Meta, Oracle
• ROCm is maturing faster than expected
• Custom silicon: Google's TPU, Amazon's Trainium, Meta's MTIA, Microsoft's Maia
• Four largest customers in the world all designing their own AI silicon

↩ rebuts: “Bull's dismissal of Intel as non-competitive due to bad margins and lack of CUDA”

Multiple compression on heroic consensus at all-time highs is a 20-30% drawdown setup—the bull admitted this

• Bull admits: 'a hot CPI reprices the multiple, not the earnings'
• A 200 bps move in growth-stock multiple from 35x to 28x on $12.89 EPS takes stock from $451 to $361
• 20% drawdown with zero change in fundamentals
• Polymarket: 86% no-cuts-in-2026
• Beta 2.22 means NVDA moves 2x the market
• If S&P pulls back 10% in hawkish-rates scenario, NVDA has 20% downside from beta alone before multiple compression
• Macro signals: sticky inflation, hawkish Fed, oil bid, geopolitical stress (Hormuz), back-to-back equity declines led by growth sectors

↩ rebuts: “Bull's 'soft CPI' base case assumption”

Technical setup shows medium-term momentum decelerating, momentum divergence at recent high, and distribution signature

• 50 SMA peaked ~$210 on June 24, now $206.25—six weeks of sideways-to-down action
• In a 'perfectly bullish' trend, the 50 SMA should be rising—it's not
• 10 EMA went from $200 to $213 in roughly two weeks (+6.7%)
• Short-term EMA accelerating into a stalled 50 SMA = momentum divergence
• RSI pulled back from 64 to 58 alongside sideways price
• Swing high of $223.96 on August 7, RSI 64.4 on that day
• Price on August 11 is $217.50, RSI is 58
• MACD May peak: +9.32
• Current MACD: +3.66 (less than 40% of prior peak)
• Volume on most recent session: 95.7M vs 130-150M on breakout days
• 200 SMA at $194 is intact, providing deep backstop

↩ rebuts: “Bull's 'perfect MA stack' bullish interpretation”

Peak buybacks are capital destruction masked as shareholder return

• Quarterly buyback: $19.3B
• Annualized buyback: $77B+
• Buying back stock within 8% of all-time highs
• Companies that have done largest buybacks at top of cycles: IBM late 1990s, Cisco 2000, Boeing late 2010s—stocks went nowhere for a decade
• Should be building inventory buffer, investing in Rubin/next-gen architecture, diversifying away from TSMC, building software/services recurring revenue instead

↩ rebuts: “Bull's 'structural bid' framing of buybacks”

Q2 FY27 earnings in 2-3 weeks is the real catalyst—the setup is asymmetric downside

• Q1 FY27 revenue: $81.62B (already priced in)
• Whisper numbers for Q2 FY27: likely $90B+
• Consensus likely $85-88B
• Any sequential growth rate below +10% QoQ will be read as deceleration
• With forward P/E of 17x and PEG of 0.6 already at the wire, low-end print triggers 10-15% drawdown
• Bull recommending buy before CPI binary event AND quarterly print in same two-week window

↩ rebuts: “Bull's buy recommendation before CPI/earnings cluster”

NVDA and its ecosystem proxies represent one highly correlated trade with no diversification benefit

• Every NVDA proxy moves on same narrative: CoreWeave beat → NVDA up, Lumentum miss → NVDA down, Super Micro warning → NVDA down, Synopsys/EDA news → NVDA up, Intel AI news → NVDA mixed, Burry short → NVDA cautious
• One trade, many names—all correlated
• 2008-2010 housing analogy: every housing-exposed name looked 'different' until they all moved together on the way down

↩ rebuts: “Bull's 'TAM keeps expanding with sovereign and neocloud diversification' thesis”