Semiconductors Defy a Mixed Tape as Energy and Software Sink
TREE NEWS reports: Nasdaq 100 constituents traded sharply divergently intraday on September 16, with semiconductor and AI-infrastructure names pushing higher while energy, enterprise software and parts of big tech slid. The split screen underscores a market rotating toward hardware and compute exposure and away from rate-sensitive and cyclical pockets.
Winners: Compute, Optics and Chip Design
Optical networking name Lumentum led the index with a 6.47% gain, followed by Intel at 5.16% and Astera Labs at 5.15%. Advanced Micro Devices rose 4.16%, Marvell Technology added 4.82%, and Nebius NV climbed 4.21%, while CoreWeave advanced 3.31%. Storage and test equipment joined in: Western Digital gained 3.52% and Teradyne 3.18%. Chip IP and EDA names were firm, with Arm up 2.82%, Synopsys 2.40% and Cadence 1.70%, alongside Monolithic Power Systems at 2.13%.
- Leaders: Lumentum +6.47%, Intel +5.16%, Astera Labs +5.15%, Marvell +4.82%, AMD +4.16%, Nebius +4.21%
- Laggards: Diamondback Energy -8.59%, Strategy -2.72%, Adobe -2.60%, Intuit -2.49%, Old Dominion -2.54%
Losers: Energy, Software and Crypto Proxy
Diamondback Energy was the worst performer, tumbling 8.59% — a move that points to sector-specific pressure on oil and gas producers rather than broad risk aversion. Enterprise software and workflow platforms were notably weak: Adobe fell 2.60%, Intuit 2.49%, Autodesk 2.08%, Workday 1.57% and Datadog — despite finishing higher — illustrates the mixed tone within software. Microsoft slipped 1.02%, Netflix 0.85%, and Palantir 1.46%.
Notably, Strategy (MSTR), the largest corporate bitcoin holder, dropped 2.72%, diverging from the semiconductor rally. That gap is worth watching: if digital-asset proxies decouple from AI-hardware momentum, it signals that the current bid is about compute capex rather than a broad liquidity wave.
Implications and What to Watch
The pattern — semis and AI infrastructure up, energy and legacy software down — is consistent with investors concentrating exposure in the physical buildout of AI: optics, memory, custom silicon and GPU cloud capacity. Names like Nebius and CoreWeave sit directly on that theme, while Lumentum and Marvell supply the interconnect layer that scales clusters.
For the week ahead, watch three things: whether energy weakness broadens into a macro signal on demand, whether enterprise software stabilizes after a soft patch, and whether crypto-linked equities like MSTR re-couple with the AI trade or continue to trade on their own liquidity cycle. A sustained divergence would suggest the market is pricing an AI capex cycle rather than a general risk-on regime.
