While Goldman runs its conference on the West Coast, Citi runs the other half of the September tech calendar in New York. The Citi Global Technology Conference, expected in its customary early-September slot in 2026, is where the hardware and semiconductor tape gets written for East Coast institutional money, and its roster has historically skewed toward the names that actually build things.
The value of Citi’s event is breadth across the supply chain. Foundry, memory, networking silicon, and the component layer all present in the same few days, which lets the buy side triangulate. When a memory supplier describes pricing and a networking-silicon vendor describes attach rates and a foundry describes utilization in the same window, the picture assembles faster than any single print allows.
Memory is the subplot to watch in 2026. The cycle that swung the group through a brutal downturn and a sharp recovery is the question the room keeps returning to: is high-bandwidth memory demand pulling the entire memory complex, or is the strength concentrated in the AI-attached portion while commodity DRAM and NAND follow their own supply discipline? Managements from across the memory cohort tend to be franker at Citi than the megacaps are at Communacopia, because the cyclical names have nowhere to hide a turn.
Networking and custom silicon are the second thread. The economics of moving data inside and between AI clusters have become a primary investment debate, and the vendors of switching silicon, optical interconnect, and custom accelerators use Citi to frame their content-per-system story. The read-through runs straight to the substrate and photonics names upstream, where a single design win can reset a multi-year revenue trajectory.
The caution at Citi is the same as anywhere: a hardware conference in a strong tape produces uniformly confident managements, and the uniformity itself is a risk. When every team describes demand as robust and visibility as improving, the cohort is priced for it, and the disappointment when it arrives is non-linear. The useful signal is the team that breaks ranks, the one that flags a specific softening or declines to extend its visibility window.
Citi will not generate the megacap headline that San Francisco does. What it generates is the granular hardware tape, the component-level detail that tells you whether the AI buildout is broadening into the full supply chain or concentrating at the top. For an investor positioned upstream, that granularity is the entire trade. The headline is set in San Francisco. The thesis is confirmed in New York.