Alkeon Capital Management 13F Portfolio · Panayotis Sparaggis

The biggest buys table makes their intent obvious: Alkeon is reloading around the picks-and-shovels of the compute cycle rather than chasing the same AI narratives everyone else is crowding into.

• AMAT (Applied Materials, new, 4.20%, $1.09B): A flagship new position at scale says they want direct exposure to wafer fab equipment volumes, not just the chip designers. With gain_vs_avg_buy_pct only 3.5%, this is capital going in close to cost, not a momentum chase.
• NOK (Nokia, new, 1.84%, $478.1M): A sizeable first-time bet on radio and network gear to carry the data deluge downstream from AI workloads.
• MKSI (MKS Instruments, new, 1.84%, $476.8M): Another critical process and subsystems supplier, notably initiated despite being slightly underwater at -7.6% vs cost — a clean signal they’re buying for structural, not tactical, reasons.
• SIMO (Silicon Motion, new, 1.41%, $366.7M): Flash controllers and storage plumbing; a direct nod to the I/O and memory bottlenecks created by AI-heavy architectures.
• NVT (nVent Electric, new, 0.80%, $206.5M): Power and electrical connectivity gear to keep data centers and industrial electrification running — an under-the-radar infra angle.
• QQQ (Invesco QQQ, +21.3%, $1.18B) and SOXX (iShares Semiconductor ETF, +19.5%, $237.9M): Growing these ETFs turns part of their AI/semis view into a liquid overlay instead of stock-picking every incremental dollar.
• APH (Amphenol, +24.1%, $1.01B) and EQIX (Equinix, +42.8%, $347.9M): More dollars into high-quality interconnects and neutral data-center REITs rounds out a coherent bandwidth-and-racks thesis.

Taken together, the adds say Alkeon is underwriting the durability of the capex cycle behind AI — fabs, interconnect, power, and neutral colo — even as they grow more agnostic on which front-end chip designer or cloud platform wins the margin pool.

The funding side is just as telling. The biggest trims read like a who’s-who of early-cycle AI and internet winners where Alkeon’s entry points were far lower than today’s marks.

• TSM (Taiwan Semi, -34.8%, $1.69B): Sitting on a 306.3% gain vs cost, this is a classic risk-budget release. They’re not bearish on foundry economics; they’re monetizing a home run.
• KLAC (KLA, -35.3%, $1.13B) and LRCX (Lam Research, -21.8%, $1.32B): Both show eye-watering gains (354.5% and 413.4%), yet are being cut back to make way for a broader semicap basket including AMAT and MKSI.
• GOOGL (Alphabet, -27.1%, $1.32B), META (Meta, -9.3%, $1.10B), and MSFT (Microsoft, -29.1%, $317.5M): They’re shaving the Big Tech platform layer after a strong run — all well over +80% vs cost — but not abandoning it.
• VRT (Vertiv, -43.2%, $476.1M) and GE Aerospace (-54.3%, $119.6M): After huge moves (Vertiv at +83.2%, GE at +123.6%), size is coming down in the most explosive infra names while they re-spread capital across EQIX, NVT, BWXT, and others.
• TER (Teradyne, -22.5%, $1.05B) and EXPE (Expedia, -26.3%, $467.1M): Test equipment and cyclical travel are being tapped as liquidity sources; both still sit on triple-digit or near-triple-digit gains.

The pattern is disciplined: harvest where multiples and market caps have expanded fastest, especially where positions have swelled via performance, and recycle into earlier-stage or less fully valued infrastructure legs of the same secular demand story.

On the sector chart, technology remains the spine of the fund at 53.43%, but that headline masks a subtle derisking beneath. Tech was 56.14% the prior quarter, and the cut mostly reflects trims in the most mature, richly valued AI and platform names rather than abandoning the theme.

Industrials are the clear gainer, rising to 15.98% from 14.37%. New positions in MKSI and NVT, plus existing stakes like TER, BWXT, VMC, MLM, PWR, TDG, and CP, give Alkeon ballast in hard-asset and engineering names tied to capex, aerospace, and infrastructure.

Real estate (properly understood here as payments and digital infra given mislabels on Visa, Mastercard, Uber, and MercadoLibre) nudges up to 7.51% from 6.97% as they add to EQIX and keep V and MA unchanged. Unclassified ETFs (QQQ and SOXX) step up to 5.73% from 4.60%, effectively functioning as a shock absorber for their tech view.

Consumer discretionary drifts slightly down to 10.74% from 11.04%, with cuts in Expedia offset by steady off-price retail (TJX, ROST, BURL) and housing-adjacent Sherwin-Williams. Utilities ease to 4.72% from 5.04% as they chip away at CEG and VST, but maintain diversified exposure via ETR and NI — a quiet hedge on power markets that rhyme with their data-center and electrification thesis.

Finance holds around 1.90% vs 1.85% with sticky stakes in CG and ICE, a low-drama earnings and fee stream that supports an otherwise growth-heavy profile.

Put together, this 2026-Q2 filing shows a manager who believes the AI and compute super-cycle is real but also recognizes that the easy money in the obvious winners has already been made. Their 3-year annualized performance of 37.77% and latest-quarter gain of 30.18% give them the luxury of playing offense from a position of strength.

The offense is clear: more exposure to the midstream of the stack. New money into AMAT, MKSI, SIMO, NVT, NOK, APH, EQIX, and the SOXX ETF says they expect a long, investment-heavy phase of building fabs, routers, racks, and power infrastructure — regardless of how AI model economics shake out. They are also visibly willing to own that view via diversified vehicles like QQQ when idiosyncratic valuations look stretched.

At the same time, sticky positions in AMZN, CDNS, and a cluster of off-price retailers and construction aggregates suggest they are not abandoning the U.S. consumer or software monetization of this cycle. They are simply capping single-name blow-up risk after a huge run.

Going forward, watch whether technology’s share continues to drift down or stabilizes around the current 50–55% zone. If semicap and electrification continue to absorb capital while platform names are trimmed, Alkeon will be signaling a long-duration bet on capacity and infrastructure, not just AI headline stories.