Sculptor Capital 13F Portfolio

The biggest dollar adds are a clean roadmap of where Sculptor thinks the next three years of AI economics accrue. Taiwan Semi is now a 6.38% position after a +299.4% share increase and an estimated $461.8M add, effectively anointing it as the core bottleneck asset in the book.

They ring-fence that view by building a broad compute basket rather than a single-name punt:

• AMD enters as a new 2.46% position, a $238.1M bet on GPU share gains and diversified AI silicon.
• Micron, at 1.67% and $161.6M, says memory bandwidth is a first-class AI constraint, not an afterthought.
• Microsoft is a new 1.64% stake worth $158.5M, framed less as “Big Tech beta” and more as an AI platform tollbooth.
• Meta, up +228.9% in shares with a $117.6M add, underlines belief that self-funded AI capex and engagement will still monetize well.
• Palantir is a fresh 1.21% position at $116.7M, a direct expression that applied AI software – not just infra – will capture economics.

Beyond pure AI, they keep scaling what look like structural growers tied to healthcare and experiences. Bruker gets a +41.4% increase (about $81.5M more), while Sotera Health, Live Nation, Viking Holdings, AerCap, and Apollo all see meaningful adds, suggesting Sculptor wants durable cash-flow compounding around their high-octane tech core rather than mere cyclicals.

On the funding side, Sculptor is not derisking AI – they’re refining it. The sharpest cuts are in chip equipment, a classic way to say “cycle is rich, rotate up the stack.”

• Applied Materials is cut -87.1% (about -$187.6M), and Lam Research is reduced -88.6% (-$172.4M), both still deeply in the green versus cost. That’s profit-taking plus a statement that edge economics are shifting from capex suppliers to the fabs and GPU vendors themselves.
• Norfolk Southern (-44.3%, -$91.9M) and EchoStar (-44.0%, -$55.8M) free up capital from traditional rails and a challenged telecom/Pay-TV bet where returns look capped or structurally pressured.
• Kenvue gets slashed -68.6% for roughly -$168.8M, a clear pullback from low-growth consumer staples in favor of higher-duration growth.

Even within AI leaders, they’re tactically right-sizing. Nvidia is trimmed -17.3% (-$25.4M) and Alphabet -11.3% (-$13.9M), both on strong gains versus cost. That looks less like a change of heart and more like risk budgeting to fund late-cycle adds into Taiwan Semi, AMD, Micron, Microsoft, and Palantir at still-attractive economics.

Smaller funding sources – modest trims in Globus Medical, Guidewire, Bank of America, and a SPAC stub (Churchill Capital XI) – round out the story: monetize liquidity in seasoned winners and lower-conviction capital structure plays to double down on clearly defined secular themes.

The sector chart makes the re-rating explicit: technology climbs to 34.8% of the book from an estimated 23.54%, while consumer, industrials, healthcare, and financials all give ground. This is not broad tech beta – it’s tightly clustered around semis and AI software.

Consumer exposure drops from 26.05% to 21.61%, and what remains is skewed toward experiences and capacity, not pantry staples. Live Nation, Liberty Live (both share classes), Viking, AerCap, Burlington, Spotify, Amazon, Allegion, and Argan paint a picture of demand for travel, entertainment, and physical access infrastructure rather than classical packaged-goods resilience.

Industrials fall from 22.25% to 16.57%, largely via Norfolk Southern and chip-equipment trims, even as they keep or build stakes in Bruker, Howmet, and Thermo Fisher – effectively upgrading from cyclicals to structural science and aero plays. Healthcare eases to 12.52% from 15.3%, but the mix leans into tools (Sotera, Bruker, Bio-Techne) and specialty providers (Globus, Brookdale), which behave more like long-duration growth than binary biotech.

Around the edges, they quietly build new infrastructure adjacencies: telecommunications appears at 1.51% via Roku and Iridium, and real estate at 0.55% via Digital Realty. That’s the start of an ecosystem – satellites, streaming distribution, and data-center REITs – around their core bet on AI compute.

Put together, Sculptor’s 2026-Q2 changes read as a clear macro-micro call: AI is transitioning from story to installed base, and the economic gravity is shifting to capacity, platforms, and data-rich applications. The move from chip tools into Taiwan Semi, AMD, Micron, and Intel, plus scaled positions in Microsoft, Meta, and Palantir, is a statement that they want to own the toll roads of the next compute cycle, not just the contractors who poured the first concrete.

The supporting cast is deliberate. Healthcare tools and services (Sotera, Bruker, Thermo Fisher, Bio-Techne) give them exposure to secular R&D and diagnostics budgets that tend to compound regardless of the macro tape. Experience- and travel-driven consumer names – Live Nation, Viking, AerCap, Liberty Live, Burlington, Amazon – express a belief in ongoing spending on experiences and convenience, not a reversion to low-vol staples.

Financials, SPAC remnants, and some legacy cyclicals are being used as an ATM: banks and rails shrink, while Apollo, Voya, and preferreds like NextEra’s issue grow modestly, suggesting a nuanced view on credit and yield rather than a top-down “rate call.” The small but notable entries into Digital Realty, Roku, and Iridium show they’re starting to connect AI’s demand for bandwidth, storage, and distribution into hard assets and satellite links.

If Sculptor is right, the next leg of AI returns won’t look like a narrow Nvidia trade; it will look like a multi-layer stack where fabs, diversified GPU players, hyperscalers, and applied-software vendors all take a slice – with healthcare tools, experiences, and infrastructure riding shotgun. If they’re wrong, this quarter will be remembered as the moment they willingly traded steady defensives and mature cyclicals for a very concentrated bet on the durability of the AI capex supercycle.