Summary
Which dominant actions and what is called “magnificent 7” actions could be the best investments? During the first phase of the frantic enthusiasm of the AI, all the Mag 7 were loved because their opportunity seemed almost unlimited. The market is now more skeptical of AI perceived winners, which led to a natural tendency to classify the group on their front prospects. We see Amazon getting in the ranks, while Nvidia continues to order first place. All names Mag 7 have something in common. They are all dominant in at least one basic competence; And they exploited the immense cash flows eroged by their basic skills to invest and develop the generative opportunity of AI. The “ranking” of MAG 7 on its intermediate duration prospects therefore implies assessing the capacity of each company to defend and feed its main activity while using this force to acquire a part in the AI ​​ecosystem. For the list of this week, we analyze and classify the MAG 7 (from top to bottom) on their apparent intermediate perspectives. We expect our rankings for MAG 7 to be fluid, but do not plan any movement at the first or last place. * NVIDIA (NVDA): We continue to put Nvidia at the top of the MAG 7 list. Unlike the other six names, the basic competence of Nvidia is artificial intelligence. The need for a realistic rendering in the GPU games, of the founding activity of Nvidia, laid the basics of the acceleration of applications and massively parallel computer which are angular stones of the formation of large language models (LLM) and to allow an inference. We classify Nvidia number one on the basis of the evidence of the company’s growth numbers, which are stellar. During its recent 4T25, a turnover of $ 39 billion increased by 78% per year. During its GTC event in March 2025, NVIDIA plans that income from data centers would exceed $ 1 billion per year in the coming years and that all the global industrial infrastructure represented an opportunity of $ 50 billions for IA renewal. NVIDIA also provides much more than its leading GPUs in industry, including its CUDA software library, and is the development of “turbo” agentic AI with models, platforms and opening partnerships. The best proof of the momentum of Nvidia is the growth of corporate infrastructure companies using Nvidia solutions to support the interest of AI. The turnover of Micron data centers has tripled from year to year and now exceeds 50% of the total business income. Micron increased the TAM for its high-band (HBM) memory opportunity to $ 35 billion in March 2025, compared to $ 20 billion about half a year ago. Among the main providers of GPU servers based in the United States, Dell Technologies posted $ 10 billion in AI separation from FY5 and provides $ 15 billion in FY6. Hewlett-Packard Enterprise left the fiscal year 1T25 with $ 8.3 billion in control of cumulative AI systems and services. The leaders in data center interconnections (DCIS), Broadcom and Marvell, both report explosive growth in these categories. * Meta Platforms (META): Meta platforms in second place, because it was among the most successful to take advantage of its basic competence (social media) in a leadership role in the construction of LLM, including multimodal models. Snap has a limited opportunity, and X is slowed down by self-inflicted injuries and the perceived contagion of musks. Only companies based in Asia such as Bytedance (Tiktok) represent a real threat. We note a risk for the core business of opposition to the apparently endless congress and the opposition of the American Congress; And Meta’s cornerstone (Facebook, Instagram and WhatsApp) could be sculpted properly in three. We do not see this happening, however. Meta has managed to swivel the costly obsession with Zuckerberg for the metavese to the opportunity for more practical and rapid development. Unlike AWS, Microsoft Azure and Google Cloud, Meta makes it possible to allow and advance the leading LLMS, even if it is not a major hybrid cloud service provider. We see Meta first and foremost an internal AI deployment to improve operational efficiency and personal targeting in order to optimize the company’s social media users, which exceeds 3.3 billion. Facebook Reality Labs, which houses Meta AI generation activities, recorded a turnover of $ 2.1 billion in 2024, or 1% of the total meta-become. Currently, it is mainly a material company (HEADESSETS VR Quest, Ray-Ban Ar Glasses), but we see an opportunity for long-term growth as LLAMA LLAMA of Meta. Just as Apple has capitalized on the large base installed iOS and targeted the individual with iCloud, Meta could offer a personalized AI offer to its three billion users. Amazon’s online retail operations and in particular its main activities are worldwide. The combined commercial exploitation margin (Americas and international) for the 4T24 was 6.1%, by far the highest of our model to return to the creation of the company. Prime media now competes with Netflix in the quality and volume of content. AWS is the main CSP on a global scale, and its growing margins and treasury flows finance the opportunity for the AI ​​generation of the company. Amazon was quick to replace the AWS chief in May 2024 when the board of directors perceived that the company was lagging behind in General AI. AWS is at an annual execution rate of $ 110 billion and the margins are at record or close heights. Both internally and in partnership with Anthropic and others, AI has produced the basic market, where customers can choose from more than 100 llms; Accelerators Trainium and Trainium 2; And Amazon Nova, a large family of fundamental models. * Apple (AAPP): Fourth place goes to Apple. Apple’s main competence is the world leadership of technological devices and services. Apple was somewhat disappointing in its first efforts in AI. Society even comes out of its usual dependence on internal capacities to obtain the support of companies such as OPENAI. But we have long considered Apple as a product perfecting, not as a pioneer of the product; Mobile phones of early iteration such as Dynatac and Startac have been surprised, but they have left for a long time while the iPhone is growing in dominance. Investors are also concerned about the drop in sales in Apple China. However, in the most recent 1T25, Apple has increased its income in all other regions. The huge base installed iOS of the company exceeds 2.2 billion, providing fertile land for continuous growth in services. * Microsoft (MSFT): in the fifth place is Microsoft. Microsoft has taken advantage of is basic competence in company software in Azure, CSP number two behind AWS. Like all the companies mentioned above which are not named Nvidia, Microsoft is challenged to obtain enough blackwell products advanced to build and manage its most advanced models. Microsoft’s cloud activities have never been so profitable as AWS, although it can be partly due to what assets are allocated to the cloud division. Microsoft had a few missteps in the deployment of its Gen Ai and Bot models
