Wednesday, November 18

Nvidia’s Challenges Preceding Jensen Huang’s Controversy

“`html

Nvidia has been a key player in the rise of artificial intelligence (AI), largely due to the vision of its recently ousted CEO, Jensen Huang. Huang took an early gamble on manufacturing chips designed for training AI models, ahead of many of his competitors.

However, just as he departs, the company faces an increasing array of competitors rapidly advancing with a host of new technologies. Under the stewardship of the newly appointed CEO, Colette Kress, Nvidia finds itself at a pivotal moment.

“Everything being sold today was designed two or three years ago,” remarked the CEO of a prominent AI firm, who spoke on the condition of anonymity to maintain commercial relationships. Kress, or her successor, will need to chart the future direction of the chip industry over the next couple of years.

Emergence of Competitors

Nvidia currently dominates the AI accelerator market, holding an impressive 80% market share, as reported by Silicon Analysis. However, AMD, its long-time rival, is making a concerted effort to produce competitive AI chips, gradually increasing its market share from a mere 2% in 2022 to between 7% and 10% by 2026.

Major U.S. cloud players such as Google, Amazon, and Microsoft are also developing more specialised chips to enhance the training and operation of AI models. Each of these companies has made significant strides, designing their own chips to cut costs and gain some independence from Nvidia.

Several startups are now offering specialised chips that promise improvements over Nvidia’s mainstream products. For example, Cerebras produces massive chips designed to power large-scale AI models, while AMD has recently acquired Taalas, a company that embeds model weights into silicon for ultra-efficient computations. However, utilizing these chips often necessitates extensive code rewrites to ensure compatibility with cutting-edge AI models.

READ:  Entera Showcases at WIRED Futures 2026: Pioneering Innovation in Functional Nutrition

Patrick Moorhead, chief analyst at MoorInStrat, a firm closely monitoring the chip sector, points out that the surge in AI agents is placing greater emphasis on inference, the process of executing already trained AI models. In this realm, Nvidia faces a diminished advantage against large cloud service providers and startups. Nonetheless, in December 2025, Nvidia acquired Groq, a startup focused on AI chip technology.

Challenges in the Global Market

While Nvidia’s fortunes have improved in the U.S., they have worsened in China. The U.S. government has been attempting to restrict access to its most advanced chips, while Beijing is making strides to develop its own chip industry. Although Nvidia is prohibited from selling its most powerful chips in China, it is still able to sell older generations. Sales of these chips, along with non-AI products, have doubled over the past year, but China now accounts for only 8% of the company’s total revenue, down from 20% in 2024.

The actual number of chips exported to other countries that may eventually end up in China or be used remotely by Chinese companies remains uncertain. Huawei, a Chinese giant that has faced U.S. sanctions for several years, has developed new types of supercomputers that utilise significantly more chips than usual to train and run AI models. Some Chinese AI companies, such as Z.ai, which produces one of the world’s leading open-source models, claim to have trained their models exclusively using Huawei chips. If these Chinese open-weight models gain traction in other countries, a market for Chinese chips may emerge elsewhere.

READ:  Helen Warren, custom black and grey realism tattoos in Hyde

The Impact of Leadership Changes

The relationships Huang maintained with China played a crucial role in his departure. He had a personal relationship with Huang Le, also known as Alice Huang, an executive at Megaspeed, a company suspected by the U.S. government of being an asset of the Chinese Ministry of State Security. Reports indicate that Megaspeed and its subsidiaries have imported Nvidia chips worth between $4.6 billion and $7.5 billion to Southeast Asia, and it is suggested that the company may have facilitated access to Nvidia chips for Chinese AI firms.

Babak Minovi, a senior analyst at Eurasia Group, a consultancy specialising in technology and geopolitics, noted that Huang was correct in asserting that cutting off access to China would benefit domestic rivals. “Huang took a pragmatic stance and is well-aware of the competition,” he explains.

Potential Consequences of Success

Nvidia may ultimately fall victim to its own success. The AI models powered by Nvidia chips are automating aspects of programming, data analysis, and scientific research that assist in chip design. Most chip design firms are now employing specialised AI models to organise components within processors, a task that traditionally required extensive engineering knowledge and expertise. Both OpenAI and Anthropic have recently highlighted the capabilities of their latest models in designing electronic components from scratch.

Moorhead suggests that the shift towards AI-driven chip design may not pose a significant threat to Nvidia, as the company is also riding the same wave. “They too have access to AI,” he notes. However, Moorhead also highlights that the rapid ability of AI models to rewrite code for different hardware types could present a challenge. For years, Nvidia has attracted developers by providing software that simplifies programming and model creation for its chips. Its Compute Unified Device Architecture (CUDA) has long been regarded as a competitive advantage, ensuring that AI companies continue to rely on Nvidia’s chips.

READ:  Eight Months Post-Social Media Ban for Minors in Australia: TikTok Usage Remains Steady

Yet, with AI models now capable of quickly transferring an AI model to a competitor’s chip, startups like Wafer, which uses AI to port models to AMD and offer them at a lower price, have seen remarkable revenue growth in recent months, according to its CEO, Emilio Andere. Wafer can have a major open-source model operational on AMD hardware at over 80% efficiency within hours, claims Andere, emphasising that AI is advancing to a point where engineers need little to no intervention in code edits or reviews.

Dan Hutchinson, a veteran chip industry analyst, suggests that Nvidia may be entering its “Intel Swan moment,” referencing the 2019 replacement of CEO Brian Krzanich, who was focused on engineering, with CFO Bob Swan. Following this change, the company’s focus shifted more towards sales than technical progress, leading to increased bureaucracy and declining revenues. “Nvidia is now on the brink of becoming an Intel,” he asserts.

“`

Leave a Reply

Your email address will not be published. Required fields are marked *