AI Is Supplying Blowout Earnings Again

Cisco reported 12% year-over-year revenue growth, 35% order growth, and raised hyperscaler AI infrastructure order expectations to $9B this fiscal year, driving the stock up 13.8%. Lumentum delivered an even stronger quarter, with revenue up 90% year over year, adjusted operating margin expanding to 32%, and guidance for 22% sequential growth; shares jumped as high as 21%. The discussion also highlighted non-AI diversification ideas including Deckers, Casella Waste Systems, Trex, Berkshire Hathaway, and Disney.

Analysis

The market is starting to re-rate AI infrastructure as a capacity bottleneck business rather than a pure growth story. That matters because the first derivative is no longer revenue surprise; it is backlog visibility and mix shift into higher-margin interconnect/optimization products, which can keep earnings revisions rising even if unit growth slows. Cisco is the lower-beta beneficiary of that re-rating: it has less upside if the cycle extends, but it also has the best chance to absorb a pause without a multiple collapse because the order book is doing the heavy lifting.

Lumentum is the more interesting second-order winner because it sits in the efficiency layer, not the brute-force compute layer. In a world where power, cooling, and throughput constraints become the real limiter, suppliers that reduce watts per bit or increase cluster utilization can compound faster than the headline GPU spend. The risk is that the valuation has already discounted several years of execution, so any evidence of capacity catching up, customer concentration, or slower-than-expected hyperscaler capex could compress the multiple faster than fundamentals deteriorate.

The contrarian read is that this is not just a generic AI trade, but a selectivity trade within AI infrastructure. Investors who treat Cisco and Lumentum as interchangeable “networking” names are missing the spread: one is increasingly a cash-flow duration asset, the other is a scarcity/optionality asset with much higher embedded expectations. If the AI capex cycle rolls over, Lumentum likely de-rates first; if it extends, Cisco likely underperforms on upside capture but outperforms on downside resilience.

Key Decisions for Investors

• Long CSCO as a lower-volatility AI infrastructure proxy for the next 6-12 months; prefer entry on any post-earnings digestion rather than chasing strength. Risk/reward: modest upside from multiple expansion, but materially better drawdown protection than higher-beta peers.
• Avoid initiating fresh full-size LITE at current levels; treat it as a trading position only. If bought, use a 3-6 month horizon and size small because the stock is pricing in a near-perfect 2026-2028 execution path.
• Pair trade: long CSCO / short a basket of high-multiple AI infrastructure beneficiaries with weaker visibility. The thesis is that order visibility and balance-sheet quality will matter more if AI capex growth decelerates over the next two quarters.
• For accounts that want AI exposure but less factor risk, rotate incremental capital into ANET over LITE on pullbacks. The cleaner balance sheet and more direct network spine exposure provide a better risk-adjusted way to own the same secular theme.
• Watch for any sign of hyperscaler order normalization or delayed facility ramps; if that appears, trim LITE first and keep CSCO as the residual holding because its earnings base is less dependent on perfect sentiment.