When: Current levels below $370

Then: Do not add until breakout confirmed

HOLD existing positions; stay flat for prospective buyers; begin 10-15% light trimming now into the bounce; binding hard stop at $340; prohibit scaling-in until weekly close above $380 with RSI above 60; cap position at 60-70% target weight; do not expand toward aggressive add-on plan.

• Forward P/E of 18.7 and PEG of 0.35 embed assumptions of near-perfect execution on VMware synergies, AI-XPU ramp, no margin compression, and no customer whipsaw; any leg of the thesis wobbles and the multiple snaps back to TTM 46x
• Price chart shows confirmed downtrend: lower-high sequence at $493, $427, and roughly $380 with price remaining under both 50 and 200 SMA; bullish MACD crossover below major moving averages has historically failed more often than it succeeds
• AR doubling from $6.5 billion to $13.7 billion may reflect pull-forward shipments, looser credit terms, or customer stress rather than genuine ramp; Anthropic's $42 billion annual cash burn makes it a binary survival customer with different credit risk profile than hyperscalers
• Anthropic's $41.97 billion GAAP losses and $42 billion annual cash burn create dependency on capital markets access; if IPO prices poorly, compute economics deteriorate, or hyperscalers tighten capex, Anthropic spend profile flexes downward
• Meta and Google publicly building in-house TPU/MTIA silicon to reduce third-party dependence; SEC filings lack visibility into contract structure or revenue concentration with these top customers
• Realistic technical downside under broken-trend scenario tests $320, then $300, then potentially $290s (52-week low $289.96); 15% drawdown assumption from $362 entry misses true range of risk
• AVGO beta of 1.48 amplifies macro moves; Polymarket pricing 96% probability of NO Fed cuts means marginal surprise is hawkish, which hits high-multiple AI names harder than other segments
• Aggressive analyst asking to scale into already 60-70% weighted position would push OVERWEIGHT a single high-beta name into a binary earnings print in an environment where concentration risk is elevated

→ vs aggressive: Forward P/E of 18.7 and PEG of 0.35 are not attractive valuations—they embed the bull case outcome rather than representing a margin of safety; the analyst is buying a name that REQUIRES a 0.35 PEG outcome to justify current price, not buying a 0.35 PEG name

→ vs aggressive: The 'three lower highs' are not healthy consolidation but a confirmed lower-high, lower-high sequence; the 50-day and 200-day moving averages are resistance levels where prior rallies failed, not support levels to add into

→ vs aggressive: AR scaling faster than revenue is mechanically an extension of days sales outstanding and a potential earnings quality warning sign, not a 'trophy' or demand confirmation; treating Anthropic's $42 billion burn rate as upside treats a binary survival company as a stable anchor

→ vs aggressive: Anthropic's IPO filing showing infrastructure spend is not a 'demand pipeline confirmation'—it reflects a company structurally dependent on continued capital markets access and hyperscaler willingness to fund compute; this is a risk, not confirmation of sustainable revenue

→ vs aggressive: Broadcom capturing more value per generation in co-design is an optimistic interpretation; the risk-aware reading is that two of the largest AVGO customers are publicly trying to internalize Broadcom's core capability—this is a strategic risk requiring margin of safety, not aggressive scaling-in

→ vs aggressive: MACD crossover with histogram flipping positive below major moving averages is a setup that historically fails more often than it succeeds until moving averages are reclaimed on a closing basis; buying a single indicator signal in a confirmed downtrend is how traders get chopped up

→ vs aggressive: The 3-to-1 risk-reward framing uses the recent swing low as stop-loss anchor when actual technical risk under broken-trend scenario tests $320, $300, then the 52-week low of $289.96; realistic downside targets must be used when discussing reward-to-risk ratios

→ vs aggressive: Polymarket's 96% 'no cuts' probability being 'already priced in' means asymmetry is skewed to the downside—the marginal surprise is the 4% hawkish tail, which hits high-multiple AI names with high beta (1.48) harder than other segments

→ vs aggressive: Scaling into a 60-70% weighted position would push OVERWEIGHT a single high-beta name into a binary earnings print in a sector at fresh highs, in an environment where concentration risk is elevated—this is not tactical entry, this is concentration risk dressed up as conviction

HOLD for existing holders with staged 20-30% trim into 3,660-3,720 resistance zone; prospective buyers stay flat and wait for pullback to 3,430-3,554 or breakout above 3,720 on volume

• Fundamentals are elite with Forward P/E of 18.7, PEG of 0.35, 54% operating margin, 44% ROE, $30B trailing FCF, but 0.35 PEG is depressed by VMware integration noise making current multiple appear cheaper than warranted
• Receivables expansion from $6.5B to $13.5B is ambiguous and requires DSO monitoring; calling it a trophy or disaster is overreach in either direction
• Anthropic represents high-beta optionality within AVGO book: incremental customer for silicon franchise but also customer whose survival depends on capital markets remaining open
• Dense resistance cluster at 3,660-3,720 zone (50 SMA at 3,723, 200 SMA at 3,663.8, Bollinger upper band at 3,713, late August swing area around $380) makes this an ideal trim zone, not the current bounce
• Momentum improving with ten EMA reclaimed at 3,554, MACD histogram positive, RSI neutral with room to run, stock up 5.5% in four sessions; trimming now is premature selling into early innings of momentum shift
• Macro environment is range-bound for rates with slight dovish skew over 6-12 months, supporting continued AI infrastructure spending; not the binary bear case the conservative analyst implies
• Customer concentration risk is real but moderate: Broadcom captures co-design and integration revenue that scales with each generation; not a thesis-breaker justifying underweight into $30B FCF name
• Hard position ceiling of 60-70% normal AVGO weight warranted given beta of 1.48, single-name concentration, binary earnings event, and fragile macro marginal surprise environment

→ vs aggressive: The 0.35 PEG is only low because TTM earnings are depressed by VMware integration noise; if synergies underperform by 15% or AI XPU pricing compresses by 10%, the 'cheap' PEG becomes 'fair' PEG in a hurry. This is asymmetric dependence on operational perfection, not asymmetric upside.

→ vs aggressive: Anthropic IPO compute spend is consistent with AI infrastructure tailwinds but a customer burning $42B/year is not the same credit risk as a hyperscaler with $100B in cash; cannot pick only the bull case.

→ vs aggressive: The aggressive plan to scale in would put us overweight into a setup where the bear case has more legs than the bull case in the near term.

→ vs conservative: Trimming 10-15% right now into the bounce is premature; just reclaimed ten EMA at 3,554, MACD histogram flipped positive, RSI neutral with room to run, stock up 5.5% in four sessions. Trimming into early innings of momentum shift is selling into strength before it expresses itself.

→ vs conservative: Trigger of weekly close above 3,800 with RSI above 60 is too restrictive: requires fully confirmed breakout before re-engagement, meaning sitting through a potential 50-70 point move before getting involved.

→ vs conservative: Macro framing is not the binary bear case implied; with recession probability only 8% and weak jobs data pushing cut expectations higher over time, risk is real but not asymmetric to the downside.

→ vs conservative: Customer concentration treating this as a thesis-breaker justifying underweight into a name with $30B FCF is overcorrecting; competitive moat is durable but not impregnable, which is a moderating factor, not a thesis-breaker.