Global financial giant UBS Group AG has disclosed its second-quarter 2025 U.S. equity holdings in a 13F filing with the SEC. As of June 30, the firm's portfolio totaled $790 billion, a significant 18% increase from $670 billion in the prior quarter, even as selling pressure mounted. In Q2, UBS added 1,281 new positions, increased 4,272, reduced 4,416, and exited 1,282. The top ten holdings account for roughly 13% of the total portfolio value.
The latest 13F filing reveals a textbook barbell strategy: a continued commitment to the structural boom in AI computing, paired with active management of tail risks from crowded trades. Rather than a one-directional chase of AI infrastructure leaders like NVIDIA and AMD, UBS is shifting from high-beta, single AI chip stocks toward storage and cloud platform giants, combined with index call options to capture upside. It is also using healthcare growth stocks and index put options to mitigate downside risks.
NVIDIA remains the top holding, but Microsoft saw a notable 8.62% increase, Broadcom rose 2.01%, and Alphabet was up 4.84%. The QQQ Call position surged 90.40%, signaling a strong upward exposure to U.S. large-cap tech and AI compute. On the protective side, the SPY Put soared 155.10%, and the Eli Lilly Call skyrocketed 4,571.11%, enhancing downside convexity and non-tech, high-quality growth exposure.
UBS boosted Microsoft by 8.62%, Alphabet by about 3.8%–4.8%, and Broadcom by 2.01%. The firm still holds roughly 6.09 million shares of Micron Technology, making it the 11th-largest position. The Micron Call position, equating to about 4.21 million shares, jumped 174.94%, while the QQQ Call position, worth about 9.69 million shares, increased 90.40%. Conversely, UBS trimmed spot positions in NVIDIA, TSMC, and Meta Platforms, while the SPY Put position expanded by about 155% and the QQQ Put by roughly 10%.
The true signal from the Q2 portfolio is not an exit from tech, but a rotation from the most crowded AI leaders toward storage giants, cloud computing platforms, and broader AI beta themes. The sharp increase in the Eli Lilly Call, combined with SPY and QQQ Puts, indicates UBS is acutely aware of potential AI leverage liquidation risks. This strategy allows the firm to bet on AI's structural prosperity through Micron calls while using healthcare growth and index puts to lower tail risk.
The bet on Micron Technology is the clearest AI fundamentals play in the portfolio. The QQQ Call acts as a cheap way to gain convexity on the AI bull market's continued upside. The surge in Micron calls reflects a belief that memory will become the second bottleneck in AI systems after GPUs. The QQQ Call allows the portfolio to capture beta and gamma from the entire AI ecosystem—including Microsoft, NVIDIA, Broadcom, and Meta—without fully concentrating capital back into the most crowded names like NVIDIA.
It is worth noting that Q2 was not an AI bear market, but there were clear signs of a correction before the July semiconductor deleveraging. On June 5 and June 23, semiconductors experienced severe selloffs, with the SOX index falling 7.9% on the latter date and Micron dropping about 13%. Despite this, the Nasdaq posted one of its best quarterly performances since 2020.
NVIDIA remains UBS's top Q2 holding, with about 83.07 million shares valued at $16.622 billion, representing 1.91% of the portfolio, a slight 0.12% reduction from the prior quarter. Apple is the second-largest holding, with about 51.8 million shares worth $14.989 billion, up 3.57% quarter-over-quarter. Microsoft is third, with about 35.88 million shares worth $13.385 billion, a significant 8.62% increase. Broadcom is fourth, with about 25.14 million shares worth $9.497 billion, up 2.01%. The SPY ETF is fifth, with about 12.27 million shares worth $9.161 billion, down 0.28%.
Positions six through ten in the top holdings are: Alphabet Class A, the Eli Lilly Call option, the SPY Put option, Amazon, and the QQQ Call option. This structure clearly shows that UBS's Q2 13F does not indicate a retreat from AI. Instead, it presents a textbook barbell configuration of "continuing to bet on the AI primary uptrend + increasing tail risk protection."
UBS held about 25.32 million shares of Alphabet Class A, worth $9.050 billion, a 4.84% increase. The Eli Lilly Call jumped to the seventh-largest position, equating to about 7.30 million shares of underlying stock, with a declared value of $8.756 billion, a staggering 4,571.11% increase. The SPY Put, now the eighth-largest holding, corresponds to about 11.29 million shares, worth $8.433 billion, up 155.10%. Amazon was held at about 34.23 million shares, worth $8.159 billion, a slight 0.51% reduction. The QQQ Call, with about 9.69 million shares worth $7.138 billion, surged 90.40% and entered the top ten.
From a portfolio change perspective, UBS's top five buys in Q2 were: the Eli Lilly Call option, the SPY Put option, Micron Technology common stock, the Micron Technology Call option, and the QQQ Call option. This configuration highlights that during Q2's high volatility and repeated deleveraging in crowded AI and semiconductor trades, UBS did not abandon the AI theme. Instead, it adopted a barbell logic: adding both Micron stock and calls, along with QQQ Calls, to strengthen bets on HBM/DRAM/NAND memory demand in the AI inference era and large-cap tech upside. Simultaneously, it massively increased SPY Puts and introduced the Eli Lilly Call to bring in healthcare growth assets with lower correlation to AI capex cycles and stronger fundamental growth, hedging against tech crowding and systemic drawdown risk.
UBS's top five sells in Q2 were: Exxon Mobil, Meta Platforms, the HYG Put option, the GLD Put option, and the GLD Call option. This is not a simple shift to full defense. By reducing exposure to traditional energy and Meta, while also lowering high-yield debt downside protection and gold option positions, UBS appears to be actively cleaning up tactical positions used for macro tail risk and commodity volatility trades. The firm is re-concentrating its portfolio around a more directional framework: "AI storage + large-cap tech upside convexity + index downside protection + healthcare quality growth to hedge AI compute selloffs."
Comments