The AI Infrastructure Stock That Could Make Investors Millionaires Is Hiding in Plain Sight, and Wall Street Isn’t Paying Attention

Broadcom reported fiscal Q3 revenue of $29.6B, up 86% year over year, and non-GAAP EPS of $3.32, nearly double from a year earlier, driven by a 221% surge in AI-chip revenue to $16.7B. Fiscal Q4 revenue guidance of $34.8B (+93% year over year) fell slightly short of the $35.1B consensus, pressuring the shares despite projected Q4 AI revenue growth of 236% to $21.7B. Management expects fiscal 2026 AI revenue of $57.6B, potentially doubling to $115B in fiscal 2027 and reaching $230B in fiscal 2028; the article argues that EPS above $30 in fiscal 2028 could support a $900 share price, or roughly 149% upside.

Analysis

The relevant debate is not whether AI spending remains elevated, but whether Broadcom can convert bespoke accelerator design wins into durable, high-margin recurring silicon content before each customer insources or dual-sources. Custom ASIC programs create multi-year qualification barriers and pull through Ethernet switching, optical connectivity and interconnect content; this makes AVGO a more direct beneficiary of hyperscaler capex mix than the market’s headline GPU-spending framework implies. The offset is extreme customer concentration: a delayed deployment at one large cloud customer can move quarterly revenue materially, making a modest guide miss more informative than a broad AI-demand slowdown.

AVGO’s relative underperformance after an earnings beat indicates the market is discounting execution and sustainability rather than current demand. If management’s long-dated AI targets are credible, consensus must raise both revenue and operating-profit assumptions over the next 1-3 months; the valuation rerating comes only after investors gain confidence that custom chips are not merely lower-margin substitutes for NVIDIA GPUs. Conversely, a rising ASIC share is strategically negative for NVDA’s accelerator share but potentially positive for NVDA’s networking franchise, limiting the purity of a simple AVGO-long/NVDA-short thesis.

The bullish retail framing is too confident on terminal multiples: at this scale, a sustained premium depends on gross-margin stability, customer prepayments/capacity commitments, and evidence that AI revenue is converting to cash rather than absorbing working capital. Watch the next earnings release for AI revenue growth versus guidance, semiconductor gross margin, inventory/receivables growth, and any customer-specific commentary. A guide reduction or a sequential margin decline alongside elevated receivables would falsify the near-term rerating case more quickly than a softer industry capex forecast.

Key Decisions for Investors

• Initiate a 1-3 month tactical long AVGO only on post-earnings weakness or after independent confirmation that FY2027 consensus AI-revenue estimates are moving higher; target a rerating versus SOX, with a 10-12% downside stop if next-quarter guidance or semiconductor gross margin disappoints.
• Express the custom-silicon-share thesis as a modest long AVGO / short NVDA dollar-neutral pair over 3-6 months, sized small because NVDA can retain networking and software economics even if accelerator mix shifts. Exit if AVGO reports a major program delay or NVDA demonstrates reaccelerating data-center revenue and stable gross margin.
• Do not underwrite the multi-year upside case until verifying customer concentration, take-or-pay/capacity commitments, and the cash-conversion profile of the ASIC ramp. Treat abnormal growth in receivables or inventory relative to revenue as an alert for a lower-quality revenue mix rather than adding exposure.
• Monitor hyperscaler capex guidance during the next reporting cycle: broad capex resilience paired with AVGO-specific weakness would signal competitive or execution risk; broad capex cuts would favor reducing both AVGO and AI-semiconductor beta through SOXX hedges.