[Why&Next] With 'Cheonvidia' Dominating... Big Tech and Anti-Envidia Alliances Also Chasing

NVIDIA Q1 Surprise Earnings
Stock Hits $1000 for the First Time

Apple Develops Own AI Chip
Intel, Qualcomm Build Software-Based Anti-NVIDIA Alliance
Market Share Expected to Drop 18% by 2030

NVIDIA, a leader in artificial intelligence (AI) chips, once again delivered a surprise performance in the first quarter of this year, exceeding market expectations. The company also announced a 10-for-1 stock split, and its stock price surpassed $1,000 for the first time in after-hours trading.

While NVIDIA's 'dominant solo leadership' remains strong, some voices have raised doubts about whether this explosive growth will continue. Following several big tech companies such as Google and Microsoft (MS), Apple has also started developing its own AI chips, signaling the beginning of a semiconductor war among big tech firms. Additionally, Intel, Qualcomm, and AMD have joined the anti-NVIDIA front, showing visible progress in software development for AI application (app) development.

AI Boom Sparks 'Earnings Surprise'... Ushering in the 'Cheonvidia' Era

On the 22nd (local time), NVIDIA announced that its first-quarter revenue surged 262% year-on-year to $26.04 billion, surpassing the market expectation of $24.65 billion. The first-quarter earnings per share were $6.12, also exceeding the forecast of $5.59. NVIDIA also announced a 10-for-1 stock split and plans to increase its quarterly cash dividend to $0.10 per share from $0.04 in the previous quarter. Following the announcement of the surprise first-quarter earnings and shareholder-friendly measures, NVIDIA's stock price soared about 7% in after-hours trading, surpassing $1,000 for the first time.

The surge in demand for generative AI across industries, including big tech, has led to a sharp increase in shipments of NVIDIA's high-performance graphics processing units (GPUs). GPUs are essential for training and inference of large language models (LLMs). Especially as the release of more powerful AI products becomes crucial, NVIDIA's data center capabilities have come into the spotlight. A data center refers to a collection of AI computers assembled by attaching high-bandwidth memory (HBM) to GPUs to create AI accelerators, along with central processing units (CPUs) and other components. NVIDIA's data center segment recorded a first-quarter revenue of $22.6 billion, a 427% increase from the same period last year.

Can NVIDIA's Solo Dominance Continue?

[Image source=Yonhap News]

However, some predict that NVIDIA's solo dominance may not last. Big tech companies are attempting to create cracks in the hardware AI chip market dominated by NVIDIA and the software development platform. It is widely accepted on Wall Street that NVIDIA's market share will gradually decline.

NVIDIA was able to dominate over 80% of the AI chip market by providing a one-stop solution that includes AI accelerators specialized for AI model training and inference, as well as platforms capable of developing generative AI. By offering software used by AI creators alongside its hardware, NVIDIA has enhanced customer convenience and expanded its market dominance.

As the global AI boom intensifies, there are ongoing observations that NVIDIA's supremacy could be challenged. Competitor semiconductor companies have launched AI accelerators, with AMD's MI300X being used primarily by Microsoft (MS), Meta Platforms, and Oracle. The five major U.S. big tech companies listed on NASDAQ, which account for a significant portion of NVIDIA's revenue, have also entered the race to secure their own AI chips.

However, the market believes that merely matching NVIDIA in AI accelerator performance is not enough. Since the release of NVIDIA's CUDA, the leading AI development software platform, in 2007, AI developers have programmed using this platform, and the accumulated code is regarded as a vast theory for subsequent developers to learn from. This has created a strong 'lock-in effect.' Gennady Pekhimenko, Associate Professor of Computer Science at the University of Toronto, explained, "To compete with NVIDIA in the AI chip field, it is necessary not only to make hardware but also to develop software that is easy for developers to use."

Anti-NVIDIA Alliance Forming in Software Market

This is why an anti-NVIDIA alliance is forming in the AI development software market. OpenAI developed the AI software Triton in 2019, and big tech companies such as Meta Platforms, MS, and Google are supporting it. Meta Platforms uses Triton for its self-developed AI chip, MTIA. Initially, Triton only operated on NVIDIA AI accelerators but has since been made compatible with Intel and AMD hardware. The UXL Foundation, an alliance developing alternatives to CUDA, includes Intel, Google, AMD, and Qualcomm. Chris Lattner, a former senior engineer at Apple and Google, emphasized alternatives to CUDA by launching Mojo, an AI programming language.

Apple is also reportedly developing software that runs on AI accelerators powered by its self-developed AI chips. As Apple CEO Tim Cook recently emphasized, "The perfect integration of hardware, software, and services is Apple's differentiating advantage in the new AI era." The synergy effect of connecting proprietary generative AI to product lines such as iPhone, iPad, and Mac without external assistance is expected to be significant, potentially impacting NVIDIA's revenue structure considerably.

"NVIDIA's Market Share to Decline to 63% by 2030"

Citigroup, a major U.S. bank, recently estimated that NVIDIA's market share in the AI chip market will decrease from about 81% next year to about 63% by 2030. However, some argue that since the AI wave is still in its early stages, centered on the tech industry, NVIDIA's scope can continue to expand. Bloomberg reported, "In the future, AI demand could expand across governments worldwide and industrial sectors such as pharmaceuticals, shipbuilding, and automotive."