Key takeaways:
- Explosive AI demand is driving a generational transfer of value to semiconductor suppliers, with NVIDIA’s scale of free cash flow generation signalling a shift in the AI value chain, facilitating semi supply security and improving of shareholder returns.
- NVIDIA’s revenue sources are now more diversified with around half of the datacentre business coming from beyond the big US hyperscalers and major internet companies, driving significant, more durable, diversified growth.
- If NVIDIA can demonstrate longer-term sustainable growth and continued innovation, as Apple did post‑2012, the market may potentially reward it despite cyclical risks.
While the market continues to be apathetic to NVIDIA’s ongoing strong quarterly earnings results, it is hard to ignore the generational free cashflow transfer from AI datacentre builders to suppliers. This is because it is enabling NVIDIA to announce one of the largest shareholder return programmes in history – with parallels to Apple’s announcement in 2012 when the market was debating the sustainability of iPhone growth, and whether Apple would be the ‘next Nokia’, which led to a valuation discount by the market at the time. Signalling conviction in more durable growth and margins, Apple was able to convince the market that it was a platform company with a durable competitive moat, which ultimately led to a valuation premium. NVIDIA hopes to embark on a similar journey; it has the balance sheet and research and development (R&D) scale to be at the forefront of recent innovation in CPUs and LPUs (Language Processing Units), while retaining a front‑of‑queue advantage in an extreme supply‑constrained semiconductor landscape.
Demand has gone parabolic. The reason is simple agentic AI has arrived. I can now do productive and valuable work. Tokens are now profitable, so model makers are in a race to produce more in the AI era. Compute capacity is revenue and profits. NVIDIA is the platform of this era.
NVIDIA CEO Jensen Huang
Main highlights from NVIDIA’s earnings call:
1. An Apple moment?
A key driver of the valuation re-rating of Apple from a S&P 500 Index discount to a premium was proving that it was not the ‘next Nokia’,1 ie. it was a platform ecosystem company with more durable growth and margins. One way Tim Cook in his early days as CEO signalled that was via their March 2012 Capital Return Programme with a US$10 billion share buyback and a quarterly dividend having not done either since 1995.2 NVIDIA has just reported almost US$50 billion in free cashflow last quarter (fiscal Q1 2027) alone. This represents a once-in-a lifetime free cashflow transfer from the hyperscalers and AI datacentre builders, to the suppliers, with NVIDIA as the leading semiconductor beneficiary. As a result, it was able to announce an incremental US$80 billion share buyback, taking the outstanding authority to US$118.5 billion. This was in addition to a 25x dividend increase to US$1 per annum. NVIDIA has a soft commitment to returning 50% of free cash flow to shareholders, subject to projected profit and free cashflow growth. That dividend has the potential to provide >1% dividend yield in the next couple of years.
2. Flexing its balance sheet strength to secure supply
The unprecedented AI demand surge has used up all the post pandemic semiconductor industry excess supply, driving shortages across multiple pockets of the semiconductor and optical supply chain. Again, taking a leaf out of Apple’s playbook, NVIDIA is taking advantage of its balance sheet strength to secure supply today, and encourage suppliers to build more capacity –total prepayments and equity stakes in key suppliers with supply commitments have swelled to circa US$145 billion. Consequently, NVIDIA is in a comfortable position in terms of supply in contrast to some peers. This is evidenced by US$13.5 billion in incremental sales sequentially last quarter, more than the total annual sales of many of its competitors.
3. Share losses overstated?
The market continues to debate NVIDIA’s long-term share of the market given the rise of competing custom silicon from Alphabet and Amazon, as well as Anthropic’s rise, relying on both of those rather than NVIDIA to train its models. NVIDIA highlighted that it is now winning Anthropic business, which could lead the company to take share in frontier models. It also sees itself as dominating the non-hyperscaler market where neoclouds, sovereigns and enterprises are much less willing to design their own compute, disaggregate workloads or experiment with other chip suppliers. NVIDIA’s business continued to accelerate with the datacentre business ex-China having grown more than 100% year-on-year, driven by its Blackwell chip and ethernet networking ramp. Despite market concerns, NVIDIA’s latest Vera Rubin chip remains on track to launch later this year, offering 35x the inference throughput and 10x the datacentre AI token revenue generation of the Blackwell.
4. Diversified growth beyond the hyperscalers
New reporting segments and disclosures were aimed at highlighting to the market that half of NVIDIA’s datacentre business today comes from beyond the big US hyperscalers and major internet companies. The rise of AI frontier models, neoclouds (or AI native clouds in NVIDIA parlance), sovereign AI as well as enterprise AI is seen as driving significant, durable, diversified growth that is less constrained by the market’s laser focus on just a handful of hyperscalers’ cashflows. There continues to be zero NVIDIA sales in China, despite the US issuing more export licenses given China’s reluctance to acquiesce to the imports.
5. #1 in AI CPUs
The rise of agentic AI has led to renewed enthusiasm for CPUs, which drive the orchestration and act as a harness for agentic AI. NVIDIA disclosed for the first time commitments for US$20 billion of standalone Vera CPU revenues this year alone, which potentially could make it the largest AI CPU supplier, ahead of Intel and AMD.
6. LPUs to remain niche
With the recent successful initial public offering (IPO) of semiconductor company Cerebras, and NVIDIA’s acquisition of Groq late last year, market recognition of language processing units (LPUs) based on low latency, high token rate SRAM (Static Random Access Memory)-based chips has increased significantly. But NVIDIA still views LPUs as relevant only for a niche section of the market and for certain inferencing workloads such as decode.
Any reference to individual companies is purely for the purpose of illustration and should not be construed as a recommendation to buy or sell or advice in relation to investment, legal or tax matters.
*All references to NVIDIA earnings information sourced from NVIDIA earnings call transcript; 20 May 2026.
1 Nokia was dominant in mobile phones in the late 1990s and early 2000s but failed to adapt to technological shifts (smartphones and software ecosystems), which led to a rapid decline.
2 Apple Newsroom; 19 March 2022.
Agentic AI: An AI system that uses sophisticated reasoning and iterative planning to autonomously solve complex, multi-step problems. Vast amounts of data from multiple data sources and third-party applications are used to independently analyse challenges, develop strategies and execute tasks.
Balance sheet strength: Refers to a company being in a strong financial position. The balance sheet is a financial statement that summarises a company’s assets, liabilities and shareholders’ equity at a particular point in time.
Buyback: Where a company buys back their own shares from the market, thereby reducing the number of shares in circulation, with a consequent increase in the value of each remaining share. It increases the stake that existing shareholders have in the company, including the amount due from any future dividend payments. It typically signals the company’s optimism about the future and a possible undervaluation of the company’s equity.
Capital expenditure (capex): Money a business spends on major, long-term assets such as property and equipment (tangible assets) or technology, software, trademarks, patents etc. (intangible assets) to facilitate new projects or investments that support business growth and expansion.
CPU: The central processing unit is the control centre that runs the machine’s operating system and apps by interpreting, processing and executing instructions from hardware and software programmes.
Cyclical risks: Risks associated with business cycles or other economic cycles adversely affecting an investment, asset class, or individual company’s profits.
Decode: In Large Language Model (LLM) serving, the decode phase is the autoregressive generation of output tokens (generating one token at a time).
Dividend yield: The income received on an investment relative to its price, expressed as a percentage. It enables comparisons of the level of income provided by different investments at a point in time.
Free cash flow: Cash that a company generates after allowing for day-to-day running expenses and capital expenditure. It can then use the cash to make purchases, pay dividends or reduce debt.
Hyperscalers: Companies that provide infrastructure for cloud, networking, and internet services at scale. Examples include Google, Microsoft, Facebook, Alibaba, and Amazon Web Services (AWS).
IPO: The process of issuing shares in a private company to the public for the first time.
Low latency: Latency is the time that elapses between a user request and the completion of that request. Low latency improves the user experience and helps applications run faster and more smoothly.
LPU (Language Processing Unit): LPU is a proprietary chip architecture developed by Groq (owned by NVIDIA). LPUs are a crucial part of LPU Inference Engines, which are a new type of end-to-end processing unit system for the applications and workloads that are most commonly associated with natural language processing or AI language applications.
Neoclouds: AI-first cloud providers offering specialised GPU infrastructure at a lower cost than hyperscalers, making AI workloads more accessible.
Re-rating: Occurs when investors are willing to pay a higher price for shares, usually in anticipation of higher future earnings or other positive fundamental catalysts.
Token: AI tokens are the fundamental building blocks of input and output that Large Language Models (LLMs) use. They are the smallest units of data used by a large language model (LLM) to process and generate text/output that is useful.
Valuation: The process of determining the fair value of an asset, investment, or firm. Among others, future earnings and other company attributes are used to arrive at a valuation.
Valuation discount/premiums: Adjustments applied to the value of an asset for example a stock price to reflect specific ownership characteristics, market conditions, or transactional dynamics. A premium increases the base value, whereas a discount decreases it.
These are the views of the author at the time of publication and may differ from the views of other individuals/teams at Janus Henderson Investors. References made to individual securities do not constitute a recommendation to buy, sell or hold any security, investment strategy or market sector, and should not be assumed to be profitable. Janus Henderson Investors, its affiliated advisor, or its employees, may have a position in the securities mentioned.
Past performance does not predict future returns. The value of an investment and the income from it can fall as well as rise and you may not get back the amount originally invested.
The information in this article does not qualify as an investment recommendation.
There is no guarantee that past trends will continue, or forecasts will be realised.
Marketing Communication.