AVGO — Broadcom Inc.

Analyst reaction was immediately positive after September 2 earnings: Cantor Fitzgerald and Rosenblatt raised targets to $600, BMO to $575, and Morgan Stanley to $505, while the consensus price target sits near $503. The only hesitation is near-term: Q4 FY2026 guidance of ~$34.8B was just below the ~$35.0B whisper number and the stock had already run hard, so shares faded modestly into the mid-$360s. Net sentiment is bullish on fundamentals, with a short-term price/positioning tug the other way.

Revenue has inflected from a VMware-assisted but otherwise moderate semiconductor grower into an AI hypergrower. YoY revenue growth accelerated from roughly 28% in Q4 FY2025 and 29% in Q1 FY2026 to roughly 48% in Q2 FY2026 and 86% in Q3 FY2026. Sequential revenue rose from $19.3B in Q1 FY2026 to $22.2B in Q2 to a record $29.6B in Q3. GAAP operating margin expanded from 36.9% a year earlier to 53.9% in Q3 FY2026; TTM gross margin is 67.7%, operating margin 48.3%, and net margin 42.9%. This is acceleration, not a quarterly blip, and management's Q4 guide of ~$34.8B implies the curve steepens further. Pre-FY2024 comparisons are distorted by the VMware acquisition, but the latest surge is organic, driven by custom AI XPUs and networking.

At 2026-09-03 pricing, the stock is not cheap: trailing GAAP P/E is 45.6x and TTM P/S is 19.6x, well above the 5-year average P/S of 13.1x. The P/E is below Broadcom's own 5-year average of 56.4x, and on consensus FY2026 EPS of $11.59 the forward P/E is about 31.7x. A reported PEG of 0.45 reflects how fast earnings are moving. I would call the stock fairly valued relative to its growth and rich relative to market multiples; it is not a value stock, but it is not priced like a company with no room for upside either.

The trailing P/E is misleading because estimates are still catching up to the AI custom-silicon ramp. At $367 the stock is pricing a strong FY2026 but only a fraction of the FY2027/28 roadmap: if AI revenue truly doubles to $115B in FY2027 and EPS moves toward the mid-to-high teens, the implied forward P/E at $367 would be only in the low 20s. The bear case is the mirror image: if AI capex pauses or one hyperscaler customer slips, a 45.6x trailing GAAP P/E leaves no cushion. This is a high-conviction growth equity, not a stable compounder at a discount.

As of 2026-09-03, RSI was 42.13 — neutral, not oversold or overbought. Price was -4.6% versus the 20-day SMA, -4.5% versus the 50-day SMA, and -0.7% versus the 200-day SMA. The stock is in a corrective/consolidation phase after a 52-week range of $289.96-$495; it is roughly 26% below the high and 27% above the low. It has come down to the neighborhood of its 200-day average without a decisive break, so there is no confirmed reversal yet — patient accumulation near support makes sense, but the chart is not flashing an all-clear oversold entry.

• Hyperscaler AI capex cycle: Broadcom's custom XPU and networking revenue depends on Google, Meta, OpenAI, and Anthropic continuing multi-year data-center build-outs.
• Data-center power and physical shell constraints: management's multi-gigawatt deployment roadmap requires access to power and co-location capacity; delays would push AI revenue into later periods.
• Interest-rate/duration environment: with beta near 1.47 and long-duration growth equity characteristics, AVGO is more sensitive to higher discount rates than the broader market.
• Semiconductor supply chain and export controls: advanced packaging, networking capacity, and US-China restrictions can affect timing, though Broadcom's hyperscaler-heavy mix partially insulates it.

• Q4 FY2026 earnings in early December 2026 — the first check on whether the ~$34.8B guidance and the FY2027/28 AI roadmap are tracking.
• New or expanded custom XPU wins with Anthropic, Google, Meta, and OpenAI, plus progress on 5-10GW incremental deployments.
• Continued margin/FCF upside — Q3 FY2026 FCF hit a record $13.7B, about 46% of revenue, which could support another re-rating if it repeats.

• Customer concentration: a few hyperscaler customers drive the AI revenue surge; a delay, cancellation, or internalization at any one would create a visible air pocket.
• Competition: merchant GPUs and in-house ASIC programs could erode Broadcom's dominant custom-silicon/networking share over time.
• Power/data-center bottlenecks could push the $115B FY2027 AI target beyond investors' time horizon.
• Valuation and positioning: 19.6x trailing sales leaves little room for estimate disappointment, and $283M+ of insider selling over three months is not a reassuring signal.

Broadcom is executing at an elite level: AI revenue grew 221% YoY in Q3 FY2026, FCF is at record levels, margins are expanding, and the company has unusually explicit hyperscaler demand through 2028. At $367.24 as of 2026-09-03, with consensus FY2026 EPS of $11.59 and likely much higher FY2027 estimates, I think the risk/reward skews positive for long-term investors who can tolerate AI capex cyclicality. This is an asymmetric growth position with real concentration and execution risk, not a low-risk value entry.

I calculated $463.52 by applying a 40x forward non-GAAP earnings multiple to consensus FY2026 EPS of $11.59 (40 x 11.588 = 463.52). The 40x multiple is above the stock's ~31.7x forward P/E but below analyst targets' implied ~43x, and I anchored it between Broadcom's 5-year average P/E of 56.4x and a PEG-1-type multiple given roughly 44% estimated EPS growth, with a discount for customer concentration and power/data-center execution risk.