Quick View: How did NVIDIA’s earnings call address AI bubble concerns?

Originally Posted 20 November 2025 – Quick View: How did NVIDIA’s earnings call address AI bubble concerns?

Portfolio Manager Richard Clode summarises the main investor implications from NVIDIA’s most recent earnings report.

Key takeaways:

• NVIDIA’s quarterly results exceeded expectations, with reaccelerating growth and incremental deals reinforcing NVIDIA’s leadership in AI infrastructure.
• The company cited generative AI, agentic AI, as well as the end of Moore’s Law as structural forces fuelling long-term demand for accelerated computing.
• Management addressed fears around GPU lifespan, circular financing, and competitive threats, emphasising strong cash flow and ecosystem expansion in an effort to debunk investor concerns.

In recent months there has been rising market angst around circular financing and an AI bubble, with hedge fund investor Michael Burry’s short thesis around hyperscalers overstating earnings via understating depreciation, circular financing and recent customer deals with competitors adding fuel to the fire.

Yesterday’s earnings call provided an opportune moment for NVIDIA to set the record straight on what the company was seeing from their vantage point as the AI leader. NVIDIA head Jensen Huang gave an impassioned defence of what is driving a huge investment in new compute infrastructure, explaining how in his view, AI still had many years of growth to come:

3 key vectors driving AI infrastructure investment

1. The end of Moore’s Law

Existing compute infrastructure that underpins all data and software no longer scales at the pace it has previously, suggesting that Moore’s Law is finally approaching obsolescence. But rather than seeing the pace of change slowing, it is instead driving a wave of innovation, pushing the need to transition to accelerated compute to reduce cost.

2. Generative AI

The internet services that we interact with daily are built on search rankings, recommender engines and advertisement targeting. Large language models (LLMs) are much more effective in driving these engines, which in turn creates higher revenue via more user engagement and traffic. LLMs are also getting better at identifying commercial intent, leading to higher click-through-rates.

3. Agentic AI/Physical AI

These new revolutionary technologies are creating brand new, very large addressable markets. This is driving some of the fastest revenue increases we have ever seen. Jensen referenced AI safety and research company Anthropic, which has seen its annualised revenue increase from US$1 billion earlier this year to US$7 billion annualised last month.¹

3 main arguments from NVIDIA’s point of view to assuage recent concerns about the company

1. Useful life of a GPU

NVIDIA’s CFO made the point that its six-year-old Ampere GPUs are still available in the cloud and fully utilised today. This contrasts with Michael Burry’s assertion that the useful life of GPUs is only two-to-three years. While a six-year-old GPU would not be used to train a highly advanced, large-scale frontier model, it can be used for less intensive workloads. The CFO also made the point that the performance of an old GPU does not sit still as software upgrades can drive meaningfully improved performance over time (4-5x for the Hopper chip over its lifetime to date).

2. Circular financing

Recently announced investments by NVIDIA in OpenAI and Anthropic need to be put in context. NVIDIA generated US$22 billion in free cashflow last quarter, it has more than US$50 billion of net cash on its balance sheet.²

NVIDIA defended these deals as a means to expand its ecosystem with Anthropic committing this week to use NVIDIA GPUs for the first time. Such deals can help accelerate the growth plans of these new customers, as well as the potential future financial returns of taking stakes in companies NVIDIA views as being generational in their capabilities.

3. Competition

NVIDIA argued that its proprietary programming language CUDA as well as versatile architecture allow the company’s GPUs to have a much longer life versus the competition, which could be limited to a few years as model technologies evolve. It is worth noting that while rival AMD at their recent analyst day promised tens of billions of dollars of AI compute revenues in 2027, NVIDIA has reported a similar amount in one quarter.³

Earnings highlights: Growth acceleration and strategic deals

In terms of earnings results, NVIDIA delivered more than the usual quantum of upside versus expectations and reaccelerating growth. Having already laid out at its recent GTC Washington event a pathway to more than US$300 billion in data centre sales next year, we saw this week’s HUMAIN and Anthropic deals as incremental to that number.⁴

The sales figures for NVIDIA continue to include de minimis China sales, given the ongoing US-China impasse. But NVIDIA has not given up hope, emphasising the need to have a competitive product to sell in China and the importance of the US being able to compete globally. Buy-side forecasts currently continue to be well ahead of formal sell-side consensus, with upgrades overnight looking to be more of a catch up.  Many investors view that NVIDIA’s stock still trades at a far from egregious valuation – this could be viewed as a strong retort to recent AI bubble concerns and comparisons with the dotcom market in 2000.