Nvidia is experiencing a very sweet moment. It has more than tripled its revenue and quadrupled net profit in just two years. And it has earned it a huge demand for its accelerators that have triggered the rise of artificial intelligence (AI). Your numbers are spectacular. In fiscal 2024, its profits grew by 581%. In 2025 they have increased by 145%. And in the 2026 fiscal year, which ended on January 25 (its fiscal calendar does not coincide with the calendar year), it had a net profit of $120,067 million.
This evolution puts an unpleasant reality on the table: this company’s business has transformed. Its engine was historically GPUs for video games, but those aren’t anymore. In the second quarter of its 2027 fiscal year ending July 26, 92.5% of its revenue came from data centers. Nothing else needs to be added. Currently, AI accelerators and their ecosystem are everything for Nvidia. or almost everything.
In this situation, Nvidia is doing what any profitable entity would: protect its market as much as possible and devote a portion of its resources to sustaining its growth. The strategy that Jensen Huang and his colleagues have developed makes a lot of sense: They are advancing billions of dollars to their suppliers to expand their facilities and reserve manufacturing capacity for their AI accelerators.
Nvidia needs to strengthen its supply chain
The demand for AI chips is growing faster than the manufacturing capacity that the companies involved in their production like TSMC, SK hynix, Samsung or Micron can support. These are the most well-known companies, but Nvidia is also needed by other corporations In fact, it has announced deals with six suppliers of components needed for its AI accelerator. These companies provide you with substrates, advanced packaging capabilities, semiconductor testing and optical engines.
Nvidia plans to increase substrate production by 2.5 times over the next three years
Substrates are the basic components for connecting AI accelerator chips to printed circuit boards. Its production is extremely complex, making its availability one of the biggest hurdles semiconductor manufacturers face in ramping up production of AI accelerators. Be that as it may, Nvidia has a plan: Intends to increase substrate production by 2.5 times in next three years. And to achieve this, it has reached agreements with Ebiden, Samsung Electro-Mechanics and Unimicron, three essential companies in its supply chain.
Advanced packaging is another fundamental asset of the Jensen Huang-led company to sustain its business. The production of an AI accelerator does not end when its chips are manufactured. Modern GPUs integrate many components, such as HBM memory, that must be assembled using very sophisticated packaging techniques. In this regard, Nvidia is supporting its leading provider of advanced packaging capabilities, Amkor, to increase its interposer wafer production capacity by more than 300% before the end of 2028. It’s certainly not lacking in ambition.
Picture | Nvidia
More information Tom’s Hardware
In Xataka | These two problems have puzzled mathematicians for decades. A genius solved them with the stroke of a pen
