SG Americas Securities 13F Portfolio

The biggest adds cluster tightly around one idea: AI is a platform transition, not a fad, and profits will compound at the platforms and in the capital equipment that feeds them. Nvidia’s debut as a new 7.16% position (about $12.5B) is the cleanest tell — they didn’t ease in; they teleported it straight into the fund’s largest line.

Megacap software and commerce were scaled aggressively as the other legs of this platform trade:

• Microsoft jumped to 5.1% after a +338.7% share increase and about $6.85B in new capital, a direct bet on AI in productivity and cloud.
• Amazon rose to 3.87% on a +367.5% share surge, a call option on AI-powered retail efficiency and AWS workloads.
• Apple, at 3.85% after a +232.5% add, is being treated less like a mature hardware name and more like a distribution gatekeeper for AI to consumers.

Down the stack, they are clearly paying for AI capacity:

• Marvell, AMD and Applied Materials saw adds of roughly $2.10B, $1.88B and $1.62B respectively, each up more than +340% in shares, while Intel and Texas Instruments were increased or initiated to anchor legacy and mixed-signal demand.
• Lam Research and KLA were scaled aggressively — KLA’s stake exploded +6697.3% in shares — signaling a belief that the bottleneck will be fabrication and process control rather than incremental model launches.
• ServiceNow, Palantir, Oracle and Alphabet’s non-voting line (GOOG) were all leaned into as enterprise and data-layer beneficiaries, suggesting SG Americas sees AI monetization diffusing across software workflows, not just GPUs.

To finance this AI escalation, SG Americas harvested gains from earlier-cycle winners and hollowed out generic beta. The most obvious ATM was Broadcom: they cut the position by -35.6%, freeing about $2.87B, even though the name still sits at 2.98% and is up 57.4% vs their average cost.

Micron is the other notable source of cash. Despite the position still being a hefty 5.75%, they trimmed shares by -18.1%, pulling out roughly $2.21B from a stake that’s now up 211.8% versus their cost basis. The message: DRAM/HBM exposure has done its job; incremental dollars now belong in CUDA and in the capital equipment cycle.

Outside semis, the moves rhyme with a “from beta and balance sheets to idiosyncratic growth” reorientation:

• SPY was slashed -57.4%, unlocking around $933.4M and signaling they see more upside (or at least better risk/reward) in names like Nvidia, Tesla and the consumer champions than in the S&P 500.
• JPMorgan and Linde were both trimmed by -15.7% and -49.5% respectively, as financials and basic materials weights fell; even Visa saw a -16% reduction, shrinking a generic payments proxy to help fund AI and consumer adds.
• Meta was shaved modestly (-7.1%), despite being slightly underwater to cost; that looks more like position sizing in an already crowded mega-cap than a thesis abandonment, especially given the simultaneous enthusiasm for Alphabet.

At the sector level, the story is subtler than “more tech.” Technology’s reported weight actually slipped from an estimated 71.47% to 66.95%, but that’s because SG Americas traded within tech — from diversified and financial-adjacent exposure toward what they see as the AI profit pool. The real rotation is into the downstream beneficiaries and shock absorbers of that same theme.

Consumer discretionary exposure more than doubled from 5.0% to 11.35%. That jump isn’t just e-commerce; it’s a composite of AI-inflected platforms (Amazon, Netflix), high-end services (Booking) and non-cyclical staples masquerading as discretionary (Costco, Walmart, Procter & Gamble) — effectively a barbell of growthy demand and quasi-defensive consumption.

Industrials moved from 1.33% to 4.6% on the back of a massive Tesla add (+559.3% in shares) plus a near-quintupling in RTX. Together with GE Aerospace, this builds an “automation and aerospace” complex that ties AI to EVs, avionics and defense budgets.

Health care climbed from 5.79% to 7.24% as they ramped Eli Lilly (+204.1% shares), UnitedHealth, AbbVie, Johnson & Johnson, Abbott and Gilead. That basket looks like a deliberate GLP‑1 and Big Pharma posture designed to offset cyclicality in tech and consumers. Meanwhile, finance, basic materials and real estate each gave up several points of share, underlining that banks, chemicals and property are now secondary in this book to AI-era demand chains.

Taken together, the 2026-Q2 moves argue that SG Americas sees AI as a multi-year earnings regime, not a crowded momentum trade to fade. They have concentrated nearly a fifth of reported equity risk in four names at the center of AI infrastructure and workloads, then ring-fenced that bet with capital equipment, EVs, aerospace and health care beneficiaries.

The trimming pattern reinforces this: they’re not bailing on semis or growth; they’re upgrading within the complex. Selling down Broadcom and Micron to fund Nvidia, Marvell, AMD, Applied Materials, Lam and KLA is a classic rotation from older winners and broad exposure into the parts of the stack with the steepest incremental demand curve.

Outside pure tech, the build-out in Tesla, RTX, GE, Costco, Walmart and the GLP‑1 complex says they expect AI-driven productivity and higher real incomes to show up in physical goods, travel, and better-funded health systems. The sharp reduction in SPY and lower bank/financial weights show conviction that security selection will matter more than macro beta from here.

Going forward, if AI capex and cloud demand keep surprising to the upside, this portfolio is set up to over-earn that trend. The key risk they are implicitly underwriting is concentration: with a 40.2% top‑10 and 66.95% in tech, any regime shift away from AI leadership would hit this book hard — but SG Americas is positioning as if that shift is nowhere on the horizon.