Key takeaways:
- NVIDIA sees 2026 as the inflection point for agentic AI, with materially higher compute intensity than prior reasoning models, supporting sustained AI infrastructure spending.
- NVIDIA reasserted leadership across compute and networking, backed by strong cash flow generation, a solid R&D budget and upcoming platform launches, reinforcing its position against rivals.
- While the market sees circular financing risks, NVIDIA’s upbeat revenue guidance and balance sheet strength make some select strategic investments in leading AI players appear reasonable.
NVIDIA’s stock has de-rated materially over the past six months; the share price has stagnated despite around 50% positive earnings revisions from the market.1 Hence the debate has been much less about stellar near-term results, and more about the sustainability of AI capital expenditure (capex) spending given concerns around its quantum, monetisation, and cash flow degradation. The muted market reaction towards NVIDIA, despite a blowout guidance mirrored last quarter, is indicative of this ongoing debate. Whether CEO Jensen Huang can move this debate forward at the GTC (NVIDIA’s annual AI global conference) event in March remains to be seen, but in the interim despite market concerns around an AI bubble and technology valuations, the shares have traded at a significant discount to its AI peers and on a not too dissimilar valuation to McDonald’s2 (which sells a different kind of chip!).
“Computing demand is growing exponentially – the agentic AI inflection point has arrived. Enterprise adoption of agents is skyrocketing. Our customers are racing to invest in AI compute- the factories powering the AI industrial revolution and their future growth.”
Jensen Huang, NVIDIA CEO
Main highlights from NVIDIA’s earnings call:
1. Agentic AI is at an inflection point
In the same way 2025 was the coming out party for reasoning models with much more intensive compute requirements and hence higher AI capex, NVIDIA views 2026 as the year of the agentic AI inflection. Agentic AI is driving higher AI infrastructure spending given the need for even more intensive compute. Addressing the concerns on sustainability of this capex, NVIDIA made the point that intelligence equals tokens; tokens are profitable for its customers and is driving higher revenues. Anthropic’s recent update is a case in point with revenues having grown 10x for each of the past three years to around a US$14 billion run rate. More recently, the agentic inflection has driven a quadrupling of business subscriptions since the start of 2026. Claude Code, launched in May 2025 is now a more than US$2.5 billion run rate business that has doubled since the beginning of 2026.3
2. Reassertion of NVIDIA’s competitive dominance
Given the recent TPU vs GPU debate, NVIDIA was keen to emphasise its clean sweep of the recent industry benchmarks (how efficiently hardware and software platforms run large language models). NVIDIA emphasised it has a US$20 billion annual research and development (R&D) budget,* and the ability to innovate across compute and networking. This should be evident with the launch of Vera Rubin, its next rack-scale platform later this year. Its networking business has grown 10x since 2021 while further innovation across the rack and the full stack will be unveiled at GTC next month.
3. Accelerating growth
NVIDIA’s revenue guidance for next quarter of around US$78 billion* was well ahead of even the most bullish market expectations, and demonstrated the fourth straight quarter of accelerating growth, which could ease concerns around a growth slowdown. Like many of its large cap peers, NVIDIA is moving to include stock-based compensation in its GAAP financials. US$35 billion of free cashflow generation last quarter* underpins the US$10 billion investment in Anthropic and an investment in OpenAI is close to being finalised. While that feeds circular financing concerns, NVIDIA’s proven strong free cash flow generation means making some select investments in leading AI players (as we saw during the internet era eg. Google and Meta appears reasonable.
4. China not factored into revenue guidance still
While the US government has granted NVIDIA licenses to ship H200s to select customers in China, and created a mechanism whereby NVIDIA pays a 25% import tariff to the US to do so,4 the company is yet to ship any H200s there. Hence we can assume no China data centre revenues in NVIDIA’s guidance given the ongoing reluctance in Beijing to accept these chips. Whether President Trump’s visit to China in March can help unlock that market remains to be seen.
AI investing needs an active approach
As a team, we continue to believe the key to harnessing the exciting investment opportunities within AI is via active investing. At present, various technological unlocks are converging to further accelerate the pace of AI innovation, including speeding up the development curve of large language models and the nearing of an inflection point in agentic AI.
*All references to NVIDIA earnings information sourced from NVIDIA earnings call transcript; 25 February 2026 and 19 November 2025.
1 FT.com; ‘Nvidia’s blockbuster results fail to dazzle investors’; 25 February 2026.
2 Bloomberg; NVDA vs MCD 12-month forward P/E as at February 2026. Past performance does not predict future returns. Forward price-to-earnings ratio is calculated by dividing the current share price by projected earnings for the next 12 months to value a company’s shares.
3 Forbes.com; ‘Anthropic Is Cashing In On Claude Code’s Success’; 17 February 2026.
4 CNBC.com; US approves Nvidia H200 chip exports to China with some conditions; 14 January 2026.
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.
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.
Circular financing: Concerns that financing of AI infrastructure investment among mega caps is becoming unsustainable. Interconnected deals and investments within a small group of companies means companies are investing in each other, with the funding recipient using the capital to make purchases from the original investor. Some of these companies may have insufficient cash flows and it could lead to a bubble when company valuations become excessive, with broader market implications.
De-rated/de-rating: Occurs when investors want to pay a lower price for shares, and assign a lower valuation to a stock, usually in anticipation of lower future earnings.
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.
GAAP: A set of generally accepted accounting principles widely used in the U.S. for financial reporting by corporations and government entities.
GPU: A graphics processing unit performs complex mathematical and geometric calculations that are necessary for graphics rendering and are also used in gaming, content creation and machine learning.
TPU vs GPU debate: This centres around specialised efficiency versus general flexibility—TPUs are custom-built to run machine learning workloads faster and more efficiently, while GPUs are more versatile and widely used across many types of computing tasks, including AI.
Price-to-earnings ratio (P/E): A popular valuation metric that measures share price compared to earnings per share for a stock or stocks in a portfolio.
Reasoning model: Learning models that make use of available information to generate predictions, make inferences and draw conclusions. It involves representing data in a form that a machine can process and understand, then applying logic to arrive at a decision.
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.
TPU: the primary task for Tensor Processing Units is matrix processing, which is a combination of multiply and accumulate operations. TPUs contain thousands of multiply-accumulators that are directly connected to each other to form a large physical matrix.
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.