AVGO Raised on Cisco Networking Supercycle

Friday, May 15, 2026

Markets extended the week's AI-led rally into the close, with SPY adding +0.79% to 748.17 and QQQ gaining +0.71% to 719.79. Beneath the headline breadth, the session's decisive action was in the semiconductor complex. Broadcom surged +5.52% to $439.79 — the largest single-session gain in the portfolio — while NVIDIA added +4.39% to $235.74, pushing its 60-day momentum to 23.24%, the highest reading in the candidate universe. But the macro picture shifted on two axes that matter as much as the price action: Fed leadership and Hormuz diplomacy.

The Networking Supercycle: AVGO's Thesis Gets Explicitly Named

Cisco's CEO used Thursday's earnings call to declare that enterprise networking is entering a "networking supercycle" driven by AI data center construction. What makes this signal tier-1 is the specificity: these are contracted purchase orders from the world's largest hyperscalers and enterprises, not aspirational AI narrative. CNBC ran a direct piece Friday titled What's good for Cisco is good for Broadcom, laying out three structural reasons why AVGO benefits in lockstep: custom ASIC contracts with Google, Meta, and Apple are scaling in lockstep with hyperscaler AI capex; the networking interconnect fabric AVGO's silicon powers is the exact layer Cisco's supercycle is constructing; and unlike speculative AI software plays, these are multi-year contracted revenue streams with embedded switching costs that do not depend on narrative continuation.

We are raising AVGO from 15% to 20%, matching NVDA's weight, to reflect this explicit demand confirmation. The ASIC lock-in thesis — our original entry rationale in April — has now been validated publicly by the dominant networking equipment vendor. Conviction score: 9.

NVIDIA continues to benefit from the same underlying capex wave. Every AI data center and networking buildout runs on NVDA GPUs at the compute layer. The +4.39% session takes NVDA to $235.74, 12.8% above SMA20 and 23.2% above SMA60. No exit signal on any measure.

The Warsh Appointment: A New Macro Risk Axis

The most consequential late-session development was not in technology. Fed Governor Miran submitted his resignation and publicly endorsed Kevin Warsh as his preferred successor to Jerome Powell. Warsh is widely understood by bond markets as a hawk — someone who believes the Fed has allowed inflation to run above mandate and would favor a more aggressive rate normalization path. The bond market's response was immediate, with yields rising and CNBC's headline framing it as the market believing the Fed is "behind the curve on inflation as Warsh takes over."

This introduces a risk axis that was not present in yesterday's portfolio construction. A hawkish Fed chair accelerates real yield normalization. That creates structural headwinds for non-yielding assets like gold and for high-multiple growth names sensitive to discount rate changes. We have adjusted the portfolio to acknowledge this uncertainty without overreacting to a leadership transition that has not yet been formally confirmed. The 20% cash reserve is the explicit buffer against this scenario materializing faster than earnings growth can absorb.

China, Hormuz, and the XLE Exit

Treasury Secretary Bessent disclosed Friday that China is working diplomatically behind the scenes to facilitate the reopening of the Strait of Hormuz. This is the most consequential geopolitical development of the week for our portfolio. Our XLE position was built entirely on the thesis that Iran's Hormuz closure would sustain a 30% OPEC production cut for an extended period, creating a durable energy supply shock and inflation premium. China's direct diplomatic intervention — with the economic leverage and geopolitical standing to meaningfully pressure all parties — materially changes the probability of a sustained closure.

We are exiting XLE. The thesis was supply-shock durability; the evidence signals that durability is now at risk. XLE's momentum picture was already the weakest among current holdings (momentum60: 0.79%), and today's +0.76% session was essentially in-line with the broad market rather than demonstrating any supply-shock premium. We close this position cleanly on thesis break rather than waiting for confirmation of what the evidence already suggests.

GLD is trimmed from 15% to 10%. The gold position faces the same Hormuz geopolitical premium erosion as XLE, compounded by the Warsh hawkish rate headwind pushing real yields higher. Gold already sits below SMA20 ($429.39) and well below SMA60 ($441.94), with a -3.33% 60-day momentum reading — the weakest technical picture in the portfolio. We are not exiting entirely: inflation re-acceleration is real, macro uncertainty has not resolved, and residual geopolitical risk remains. But we reduce concentration to reflect the dual headwind environment. At 10%, GLD is a partial hedge, not a conviction long.

AI Froth: Real Demand and Speculative Excess Are Running Together

Cerebras Systems debuted on the Nasdaq Friday at +68%, pushing its market cap to $95 billion. This is an AI chipmaker without the hyperscaler contract coverage of NVDA or AVGO. In the same session, Ford Motor surged nearly 20% over two days on an "AI energy pivot" narrative CNBC described explicitly as meme-stock dynamics — Ford's core automotive business is unchanged, but AI association has made it a momentum vehicle for retail speculation.

Goldman Sachs described the current technology surge as an "up crash" — their framework for momentum-driven rallies that overshoot fundamentals before violent reversal. Short sellers are publicly building positions against what they identify as "fake AI stocks." These signals, taken together, tell us the AI trade is entering a speculative phase alongside genuine demand. Our response is to concentrate in names with the clearest direct AI revenue confirmation (NVDA, AVGO) while maintaining 20% cash as explicit optionality against momentum reversal.

XLK is raised from 5% to 10% to capture technology sector breadth efficiently. When the AI supercycle is genuine — as confirmed hyperscaler order books now demonstrate — the entire technology infrastructure sector participates. XLK at 10% provides broad-sector leadership exposure without adding single-name concentration at a moment when AI froth signals are building in secondary names.

Institutional Context

Scion Asset Management (Michael Burry) holds NVDA at 13.5% of a concentrated, high-conviction book. Bridgewater holds NVDA at 2.6% within a broadly diversified institutional framework. Both represent validators of real AI GPU demand. Pershing Square (Bill Ackman) holds AMZN at 14.3% — the same hyperscaler capex cycle thesis that underpins our AWS conviction. Berkshire Hathaway holds CVX at 7.2% for long-duration energy exposure, but Buffett's energy thesis is company-specific and multi-decade in horizon — it does not validate the Hormuz supply-shock thesis that drove our XLE position, and Berkshire's holding does not change our exit rationale.

What Could Break the Thesis

Warsh's hawkish stance, if it accelerates rate normalization faster than AI earnings can absorb, is the primary portfolio risk. Goldman's "up crash" framing is not alarmist: high-momentum trades in semiconductor names can unwind faster than they build when institutional positioning becomes crowded and retail speculation joins in force — and both conditions are now present. A failure of China's Hormuz mediation would partially invalidate our XLE exit as mistimed, though the thesis-break signal was genuine on the available evidence and we do not re-enter on hope of a binary diplomatic failure. The 20% cash reserve is not a rounding bucket or a failure of conviction — it is an intentional allocation against a macro environment where a new Fed chair, geopolitical resolution risk, and speculative AI excess are all live simultaneously.