Susquehanna International Group 13F Portfolio

The biggest buy list is almost a checklist of who will own AI economics if this cycle persists. Susquehanna added roughly $2.96B to SPY, but the more revealing moves are single‑name and sector ETFs tied directly to AI and high‑end compute.

On the core AI infrastructure side:

• NVDA: Shares up 135.2%, lifting it to 3.70% of the book with a hefty $4.16B position and a 63.5% gain vs their average cost, shows they are adding into strength rather than top‑ticking a trade.
• AVGO, TSM, and INTC: Broadcom up 102.9%, TSM up 26.9%, and Intel up 12.5% point to a bet on connectivity, accelerators, and foundry capacity as structural, not cyclical.
• SMH and SOXX: Increases of 241.9% and 60.5% respectively say they still want semis beta, but in diversified form.

Then come the AI platforms and data monopolies:

• MSFT and GOOG: Microsoft (+70.1%) and Alphabet GOOG (+115.8%) are classic picks for AI productivity and cloud monetization, not just search and Office.
• META and CRM: Meta (+130.5%) and Salesforce (+606.0%) are high‑beta ways to play AI‑driven ad targeting and enterprise software, even though Meta is roughly flat vs cost and CRM only modestly above.
• PLTR and SHOP: Palantir (+405.7%) and Shopify (+69.2%) are smaller but telling adds — bets that AI will deepen moats in data analytics and e‑commerce enablement.

Finally, they’re clearly leaning into AI’s downstream demand via consumer and travel:

• AMZN and NFLX: Amazon (+134.1%) and Netflix (+94.2%) tie AI to consumption, logistics, and engagement.
• COST and BKNG: Big percentage increases in Costco (+143.9%) and Booking (+140.5%) show confidence that affluent consumer and travel budgets will keep flowing even as rates bite.

The through‑line: if AI drives productivity, data intensity, and discretionary income, Susquehanna wants exposure at every layer from chips to checkout.

The trims read like a classic “sell what worked, fund what’s next” playbook. The biggest dollar reductions cluster in high‑beta semis and defensive ballast.

On the chip side, they are recycling spectacular gains:

• SNDK and MU: SanDisk is cut 67.4% and Micron 21.5%; both sit on enormous gains vs average cost (345.2% and 310.1%). These look like disciplined profit takes after a ferocious memory cycle.
• AMD and MRVL: Advanced Micro Devices is reduced 54.2% and Marvell 23.4%, despite being up 198.4% and 182.4% vs cost. That’s not a thesis break, it’s risk management: rotate from more speculative CPU/accelerator plays into broader AI plumbing (NVDA, AVGO) and diversified ETFs.
• ASML and WDC: Mild trims in ASML (-5.3%) and Western Digital (-9.3%) fit the same story — lighten capital‑equipment and storage cyclicals after big runs.

Outside tech, they raided prior “safety” trades:

• UNH and LLY: UnitedHealth is down 40.8%, even with gains still positive, while Eli Lilly is increased sharply (+276.2%). That mix suggests they’re exiting managed‑care defensiveness to concentrate what healthcare they keep in the highest‑growth obesity/diabetes franchise.
• TSLA and WMT: Tesla (-25.6%) and Walmart (-24.6%) are trimmed, likely funding AMZN, COST, and BKNG — swapping mature consumer winners for fresher growth and travel leverage.
• LQD: Investment‑grade credit exposure drops 27.7%, while they add TLT (+26.2%). That’s a subtle shift from spread product toward duration, consistent with positioning for easier policy or a slower macro backdrop.

Even within broad tech beta, QQQ is cut 30.8% while SPY and SOXX are boosted. Susquehanna is saying it doesn’t need to pay pure large‑cap growth multiples for every dollar of upside when it can custom‑build its AI basket and hold cheaper broad beta around it.

Sector‑wise, the portfolio is now unapologetically tech‑centric. Technology rises to 52.54% from 49.81%, while unclassified ETFs (mostly equity and bond indices) slip from 37.36% to 34.61% — more of the performance is being driven by idiosyncratic tech risk, not just market exposure.

Within that tech slab, there’s an internal pivot from pure semis to a more balanced AI ecosystem. Heavy adds in MSFT, GOOG/GOOGL, META, ORCL, NOW, CRM, SHOP, PLTR, and QCOM mean software, cloud, enterprise workflows, and networking now shoulder more of the thesis alongside NVDA, AVGO, TSM, and the semi ETFs.

The most notable rotation outside tech is into Consumer Discretionary, which jumps to 5.62% from 3.72%. The increased stakes in AMZN, NFLX, COST, and BKNG — partially funded by cuts in WMT and TSLA — show a deliberate skew toward higher‑growth, higher‑operating‑leverage names.

Health Care and Industrials both shrink: Health Care falls to 1.83% from 2.85%, and Industrials to 2.95% from 4.42%. That’s classic risk‑on behavior — reduce defensive and cyclical exposure (UNH, some TSLA) to chase structural growth themes.

Smaller but telling moves: Energy rises to 0.65% from 0.29% on a 177.9% increase in Bloom Energy, and Real Estate climbs to 0.50% from 0.18% via UBER (+257.4%), which is mis‑classified but economically a mobility/tech name. Finance edges down as GS is essentially held steady, showing no big bet on financials as AI’s main winners.

Taken together, Q2 positioning suggests Susquehanna believes the AI boom is evolving into the market’s baseline rather than a speculative corner. The mix of bigger SPY exposure, more semis ETFs, and aggressive adds in AI platforms says they expect the winners to be broad, compounding, and index‑defining.

The profit‑taking pattern matters: they’re not abandoning chips; they’re upgrading within the theme. Trimming AMD, MU, MRVL, and SNDK to feed NVDA, AVGO, equipment names like KLAC/LRCX/AMAT, and software beneficiaries implies a view that capex and software monetization will take over from first‑wave GPU scarcity as the next leg.

The renewed push into premium consumer, travel, and select clean‑energy via AMZN, COST, BKNG, NFLX, and BE hints at a macro base case of resilient demand even if growth slows — especially for higher‑income households. Cuts in LQD and adds in TLT further suggest they see more upside in rates normalizing than in credit spreads compressing.

If this read is right, future quarters should show more of the same: tech remaining above 50% of the book, with continued shifts from narrow AI “trades” into diversified AI “infrastructure plus platforms” baskets. Any real change in that pattern — cutting NVDA, MSFT, or the semi ETFs in size, or rebuilding healthcare/credit defensives — would be the early tell that Susquehanna thinks the AI super‑cycle is finally maturing.