NVIDIA CEO Orders Extra Wafers from TSMC for AI Chips
Move counters AI bubble concerns as demand for Blackwell GPUs surges
NVIDIA, the leading AI semiconductor company, has placed additional wafer orders with Taiwan Semiconductor Manufacturing Co. (TSMC) to meet surging demand for its graphics processing units (GPUs). On the 8th (local time), NVIDIA CEO Jensen Huang stated, “We are experiencing extremely strong demand for our latest ‘Blackwell’ GPUs,” and announced that the company has requested extra wafers from TSMC. This move directly counters market concerns about an “AI bubble,” as NVIDIA is ramping up AI chip production.
Despite ongoing debates about an AI bubble, tech companies remain optimistic, expanding production and investments. Some analysts have recently raised concerns that AI technology’s value and demand may be overestimated. Last week (3rd–7th), the Nasdaq Composite Index, which includes many tech firms, fell by 3%, marking its largest drop since the announcement of reciprocal tariffs in April last year. However, tech companies continue to scale up investments and production, confident in the sustained demand for AI and its underlying semiconductors. They argue that AI is driving a genuine industrial shift, not just a speculative bubble.
◇Jensen Huang: “AI Chip Demand Is Exploding”
After attending TSMC’s sports day in Taiwan on the 8th, Jensen Huang told reporters that NVIDIA had requested additional wafers from TSMC. Wafers are the silicon substrates used to fabricate semiconductor chips. NVIDIA outsources all its AI chip production to TSMC, and the latest request aims to increase output for Blackwell and other advanced chips. Huang explained, “While NVIDIA produces GPUs, we also manufacture central processing units (CPUs), network equipment, and switches, resulting in a vast array of chips related to Blackwell,” adding that wafer demand has risen accordingly. TSMC Chairman Che-Chia Wei confirmed the request but did not disclose specific quantities. NVIDIA is known to produce 4–5 million data-center GPUs annually, but current demand far exceeds this capacity.
Production of high-bandwidth memory (HBM) for NVIDIA’s AI chips is also surging.
Huang noted,
SK Hynix, Samsung Electronics, and Micron have massively expanded their production capabilities to support us,”
And added,
NVIDIA has received samples of the most advanced chips from all three memory manufacturers.
It is understood that Samsung and SK Hynix have significantly scaled production to meet NVIDIA’s soaring chip demand. The companies also plan to supply sixth-generation HBM (HBM4) for NVIDIA’s next-generation chip, “Rubin,” set for release next year.
In a recent Bloomberg TV interview last month, Huang dismissed AI bubble concerns, stating, “I don’t believe we’re in an AI bubble. The various AI models and services we use have real value, and we are willing to pay for them.”
Big Tech Continues AI Investments
Other major tech firms are also pressing ahead with AI investments. On the 7th, Meta announced a $600 billion (approximately 880 trillion Korean won) investment over three years in the U.S. AI industry. This pledge was made by CEO Mark Zuckerberg during a meeting with U.S. President Donald Trump at the White House in September.
Meta stated,
We are investing to build world-class AI data centers in the U.S.,
Adding,
This isn’t just about powering technology and AI workloads—it’s about driving economic growth and supporting jobs and businesses nationwide.
Meta also revealed that it has added 15 GW (gigawatts) of new power to the grid through direct investments.
OpenAI, which is spending up to $500 billion on its “Stargate” project, has urged the White House to expand tax credits for AI infrastructure. In a letter, OpenAI argued that the scope of tax credits under the U.S. CHIPS Act should extend beyond semiconductor manufacturing to include the entire supply chain. The company advocates applying the 35% tax credit provided by the CHIPS Act to AI data center construction and AI server manufacturers. Despite currently operating at a loss, OpenAI aims to leverage government support to advance its AI infrastructure investments.
READ the latest news shaping the AI Chips market at AI Chips News
NVIDIA CEO Orders Extra Wafers from TSMC for AI Chips, source
Add comment