The current AI boom is increasingly fragmenting into several interconnected layers, from financial schemes and IPOs to a profound restructuring of data center hardware. On the surface, this appears as a series of high-profile deals and record-breaking IPOs, but underlying this move is a far more fundamental shift. Control over computing infrastructure is becoming a key strategic asset.
In China, this process is currently best seen through the IPO market. The government’s import substitution policy has created near-perfect conditions for AI chip developers to go public. MetaX and Moore Threads have seen their stock prices skyrocket by hundreds of percent since their IPOs, despite their lack of profitability and modest market shares.
Investors are effectively betting not on current financial performance, but on politically guaranteed demand. This optimism is partially reflected in global markets, as nasdaq futures are sensitive to such developments. Growing interest in AI chips and infrastructure is boosting capital inflows into the tech sector as a whole, albeit with a noticeable hint of speculation. However, the risk of overvaluation is obvious, as competition among dozens of players and dependence on subsidies can quickly dampen enthusiasm.
On the other side of the global market, a much more sophisticated financial structure is emerging. OpenAI is gradually becoming the center of a system where cloud giants act simultaneously as investors, infrastructure providers, and strategic partners. Amazon’s potential participation, whether through investments of up to $10 billion or the supply of Trainium accelerators, only reinforces this model. OpenAI, meanwhile, bears virtually no risk, as partners essentially bear capital expenditures and technological experimentation, while the startup itself accumulates intellectual and market power. Discussing a future market capitalization of $500 billion or even $1 trillion in the event of an IPO may seem bold. Still, it is logical given OpenAI’s significant role in the AI ecosystem.
At the same time, a less visible but no less critical hardware race is underway. SK hynix and NVIDIA Storage Next’s joint project demonstrates that the increasing challenge for AI platforms is not computation, but rather input/output and data handling. The transition to PCIe 6.0 and the leap in IOPS by an order of magnitude or more represent an effort to bridge the gap between GPU performance and data storage capabilities, particularly in the era of inference. In fact, the market is realizing that HBM alone is no longer sufficient, and next-generation NAND memory is emerging as a full-fledged component of AI architecture, rather than an auxiliary one.
Apple’s strategy fits into this logic. The company traditionally enters new markets with the help of partners and then gradually takes control of key technologies. A potential collaboration on specialized AI chips for its own data centers with Broadcom, a major player listed in the top 10 on the stock screener by market capitalization, is a typical intermediate stage. Unlike cloud giants, Apple is not expanding its infrastructure explosively; instead, it proceeds cautiously, laying the foundation for Apple Intelligence services by 2027. Meanwhile, Foxconn remains a key link across the entire value chain, from consumer electronics to AI servers, which only underscores the scale of its business transformation.
From a broader perspective, it becomes clear. The AI market is entering a phase where growth is no longer determined solely by algorithms and models, but by control over capital, production, and supply chains. Chinese IPOs, OpenAI’s financial orbit, the race for ultra-fast SSDs, and Apple’s cautious but systemic expansion are all different manifestations of the same process. In the short term, it fuels asset revaluations and increased volatility, including technology indices. Still, in the long term, it creates a new hierarchy of players, where the winners will not be the loudest, but those who manage to link technology, money, and infrastructure into a single and sustainable system.
Sandra Larson is a writer with the personal blog at ElizabethanAuthor and an academic coach for students. Her main sphere of professional interest is the connection between AI and modern study techniques. Sandra believes that digital tools are a way to a better future in the education system.
