Core Viewpoint - AI stocks are expected to continue their upward trajectory in 2026, presenting significant investment opportunities as the technology is still in its early stages [1] Group 1: Nvidia - Nvidia holds over 90% market share in GPUs, essential for AI workloads, and is poised for growth with increasing data center spending and recent approval to sell H200 chips to China [3][5] - The AI data center market is projected to grow to between $3 trillion and $4 trillion by 2030, positioning Nvidia to capture a substantial share of this opportunity [5] Group 2: Broadcom - Broadcom is emerging as a competitor to Nvidia by designing custom AI ASICs, which are energy-efficient and tailored for specific tasks, crucial as the market shifts towards inference [6][8] - Broadcom's early custom-AI chip customers represent a market opportunity exceeding $60 billion in fiscal 2027, with significant orders from Anthropic and other clients [8][9] Group 3: Alphabet - Alphabet has a significant advantage in the AI space with its seventh-generation tensor processing units (TPUs) and a leading large language model, Gemini, enhancing its cost efficiency in AI model training and cloud computing [10][11] - Alphabet is the only company with proven custom AI chips deployed at scale, making it a strong contender in the AI market [11] Group 4: Taiwan Semiconductor Manufacturing - Taiwan Semiconductor Manufacturing Company (TSMC) is crucial for chipmakers, providing advanced chip manufacturing with high yield rates, making it a preferred foundry partner [13] - TSMC anticipates a mid-40% compound annual growth rate in demand for AI chips and is expanding capacity to meet this demand, alongside expected price increases in 2026 [14][15]

Core Viewpoint - Nvidia and TSMC are highlighted as leading tech stocks to consider for investment as the market momentum continues into 2026, driven by strong performance in the tech sector [1] Nvidia - Nvidia maintains over 90% market share in the GPU data center space, solidifying its position as a leader in AI infrastructure despite facing increased competition [2] - The main challenge for Nvidia comes from custom AI ASICs, which are more energy-efficient but lack the flexibility of GPUs [4] - Nvidia's GPUs are readily available at scale, and the company collaborates closely with TSMC to secure future supply to meet rising demand [6] - Nvidia predicts that data center capital expenditures could reach $4 trillion by 2030, positioning the company to capture significant market share [7] - The stock is attractively priced with a forward P/E ratio under 24.5 times 2026 estimates and a PEG ratio below 0.7, indicating potential undervaluation [7] Taiwan Semiconductor Manufacturing (TSMC) - TSMC is the largest semiconductor contract manufacturer globally and benefits from the struggles of its rivals, being part of an oligopoly with Samsung and Intel [8] - TSMC has consistently produced chips at small node sizes with low defect rates, with 60% of its revenue coming from chips at 5nm nodes or below [10] - The company is expanding capacity by building new fabs and is expected to raise prices next year due to strong pricing power [11] - TSMC is set to introduce its new 2nm node technology, projected to cost 50% more than its 3nm technology, contributing to future growth [11] - The stock is also attractively valued with a forward P/E of 24 times, making it a solid buy given its growth outlook [12]

Group 1: Duquesne Family Office's Investment Strategy - Duquesne Family Office, led by billionaire investor Stanley Druckenmiller, has exited its stake in Palantir Technologies, a prominent AI stock, and is focusing on another critical company in the AI sector [1][5][14] - The firm held nearly 770,000 shares of Palantir as of June 30, 2024, but completely sold its position by March 31, 2025 [6][8] - The average holding period for the 69 securities in Duquesne's portfolio is less than seven months, indicating a strategy of profit-taking after significant gains [9] Group 2: Palantir Technologies - Palantir Technologies is known for its AI and machine learning platforms, Gotham and Foundry, which have secured multiyear contracts with government entities and businesses [7] - Despite its strong performance, Palantir's price-to-sales (P/S) ratio is significantly high at 115, far exceeding historical norms for leading companies in emerging investment trends [11] - Since going public in September 2020, insiders have sold over $7.6 billion in stock, with only one insider purchase, raising concerns about the stock's attractiveness [12] Group 3: Broadcom's Investment Appeal - Duquesne has re-entered a position in Broadcom, purchasing over 86,000 shares worth approximately $24 million as of mid-2025 [15] - Broadcom's hardware is essential for AI development, connecting numerous graphics processing units in data centers to enhance compute capacity and reduce latency [17] - The company is projected to generate $60 billion to $90 billion in revenue from custom AI application-specific integrated circuits (ASICs) by 2027 [18] - Broadcom's diverse revenue streams from wireless chips, cybersecurity, and industrial solutions provide stability in case of an AI market downturn [19] - The valuation of Broadcom is attractive, with a forward price-to-earnings ratio of less than 20, while expected sales growth exceeds 20% annually [20]