To The Moon

On Wednesday, following the conclusion of the GTC Berlin conference, NVIDIA (ASX: NVDA) CEO Jensen Huang and Silicon Valley tech investor Gavin Baker sat down for an interview with Yahoo Finance, during which Huang directly addressed three core issues that have recently weighed on AI investors: the debt financing risks facing new cloud computing companies, the actual useful lifespan of GPU assets, and the striking inference gross margin figures disclosed by DeepSeek.

Regarding debt financing risks and GPU lifespan, Huang made clear that NVIDIA GPUs can last seven to eight years, far exceeding the four years previously assumed by the market, a characteristic that makes them viable as loan collateral. He also noted that NVIDIA has provided guarantees on the residual resale value of its chips in certain financing projects to help lower borrowing costs for new cloud companies such as CoreWeave.

In addition, Gavin Baker cited recently disclosed data from DeepSeek showing that the open-source model's GPU inference monetization price reaches as high as $38 per hour, far above current market rental rates, which may explain why Blackwell chip rental prices have risen for 17 consecutive trading days. Huang responded by saying that strong demand is the fundamental prerequisite for solving all current financing and supply issues, adding that "with demand, all problems can be solved."

These remarks carry direct implications for the market. New cloud computing companies such as CoreWeave (NeoCloud) have recently triggered investor concerns over their highly leveraged financing structures, with their share prices and bond prices under sustained pressure. Huang's response attempts to reshape the market's understanding of AI infrastructure debt risk from the perspective of asset quality, while DeepSeek's inference margin data provides new corroboration for the sustainability of GPU demand.

Reassessing GPU Depreciation: From Four Years to Seven or Eight

Huang placed heavy emphasis during the interview on the durability of NVIDIA's GPU assets, framing it as one of the core solutions to the current AI financing predicament.

He stated that the market had widely used four years as the GPU depreciation cycle, but that this assumption is now outdated. "People are discovering that what was thought to be a four-year lifespan may actually be eight years or even longer." He attributed this characteristic to NVIDIA's continuous upgrades to the CUDA software stack, which keeps hardware productive throughout its entire lifecycle. He also confirmed that the next-generation chip, Vera Rubin, is likewise expected to have a useful life of seven to eight years.

Huang further elaborated on what this means for capital markets: the high residual value of GPUs, their broad applicability (capable of supporting all AI models and cloud service providers), and their software programmability together form the basis of collateral that lenders can underwrite. "Most GPUs today have almost no residual value, but people now recognize that NVIDIA's computing assets carry quite high residual value, and therefore can be used to back the debt itself." He stated that under this logic, NVIDIA's capacity to bear AI infrastructure financing risk is actually superior to any other asset class.

DSX Pod: Compressing the Gap Period and Accelerating Capital Returns

Addressing the core financing pain point for new cloud companies like CoreWeave — the months-long gap between equipment procurement and revenue generation — Huang introduced NVIDIA's newly launched DSX Pod solution, a modular deployment format that pre-builds AI factories inside shipping containers.

He explained that building a gigawatt-scale data center currently costs roughly $50 billion, and that from equipment leaving the factory to completed deployment and operational launch typically takes four to six months. During this period, capital sits idle while interest costs have already begun accruing. DSX Pod, by moving part of the manufacturing process forward into the supply chain, is expected to compress this cycle to three to four months, or even shorter.

Gavin Baker agreed, noting that against the backdrop of $1 trillion in AI capital expenditure, a deployment cycle of six to nine months would mean hundreds of billions of dollars in capital sitting unproductive for extended periods, creating systemic pressure on financing structures. Huang said, "If we can shorten the time from system shipment to deployment and operation from months to weeks or even one month, that would be a complete game changer for all AI factories."

DeepSeek Inference Margin Data Draws Attention

During the interview, Gavin Baker cited newly disclosed operational data from DeepSeek to offer a new quantitative perspective on the current strength of GPU demand.

According to Baker, DeepSeek has publicly stated that its open-source model's GPU inference monetization price reaches $38 per hour, while prevailing GPU rental prices in the current market are far below that level. He believes this data may explain why Blackwell chip rental prices have risen for 17 consecutive trading days. Huang responded by saying that strong demand-side momentum is the fundamental prerequisite for resolving all current financing and supply issues. He cited robust demand from multiple organizations including OpenAI, Anthropic, Meta, xAI, Google Gemini, DeepSeek, and Moonshot AI, and stated: "Without demand, no problem can be solved; with demand, all problems can be solved." He also characterized the current moment as a critical inflection point at which AI is officially switching from an "invention mode" to a "practical mode," arguing that the continued rise in token demand has solid fundamental support.

China's AI Development and the AI Employment Narrative

In addition, Huang stated that China is the only country in the world capable of achieving large-scale output simultaneously across three dimensions: chip manufacturing, open-source model development, and AI researcher cultivation. "No other country can cultivate AI researchers on the scale that China does." He also said that the United States' advantage lies in possessing the world's top frontier laboratories, chips, and computing infrastructure, and that NVIDIA will continue to compete actively worldwide, including in emerging markets such as Africa, Central Asia, Brazil, and Egypt.

On the issue of AI and employment, Huang directly characterized the prevailing narrative as "completely wrong." Citing employment growth data from sectors including energy, chips, infrastructure, construction, and healthcare, he argued that AI is actually creating more jobs at every level rather than eliminating them. "The narrative on employment is almost entirely backwards — it is completely wrong." He also noted that AI data center builders should communicate in advance and proactively integrate when entering communities to improve public perception, and said the industry as a whole has already made progress in that direction.