To The Moon

Nvidia Inc's (ticker: NVDA) second quarter (Q2) results for Fiscal Year (FY) 2027 came in at roughly a 5-10% surprise. The market’s immediate reaction in the after-hours session could best be described as ambivalent.

This caution is certainly merited in light of developments and trends.

Trend Analysis

As of the first half (H1) of FY27, Nvidia’s bottom-line merits mention:

In H1 2026, the company accrued revenue equivalent to 82% of that accrued throughout the entirety of its FY26 with cost of revenue trending at running at least 42%, which is 15% lower than the growth seen in the previous FY. Stock-Based Compensation is trending towards a 24% growth over the previous FY which had shown a 35% growth over FY25. Both Net Income Per Share and GAAP EPS (earnings per share, diluted) already stands nearly at par with entirety of FY26 – so the results are, on its face, highly encouraging.

In segment mix, corporate spends continues to grow unabated while the “Graphics” segment – arguably the more retail consumer-oriented of the two – wanes into progressively lower impact.

This helps explain the substantive bottom-line expansive: corporate products simultaneously have higher margins and create more resilient lock-in via the software ecosystem – CUDA, et al – the products are packaged with.

As the company’s client mix consolidated, so did its inventory:

The nearest period, the inventory growth has been at this level was in the post-pandemic ramp-up seen in FY22 – which was the beginning of the commercialization of AI and the beginning of the “AI Hype”. With client forecasts being stable, this reading arguably could be taken to imply a level of predictability in sales. 

Nvidia’s guidance for Q3 of FY27 has seen a modest revision upwards in the earnings release, based mostly on volume shipments of Blackwell Ultra infrastructure, which drove record server average selling prices (ASPs) in Q2 2027 and is likely to power forward into the rest of FY27. In Q3 FY2027, gross margin guidance is set at 74.0% (±50 bps) – a slight step down from 75% and reflective of the higher mechanical and thermal complexity of scaling liquid-cooled, full-rack systems being integrated.

In terms of geographic revenues, Nvidia is definitively bound to selling mostly to one region for sales growth: the US.

The US has risen from a mere 8% of net revenue share at the turn of the decade to 70% in H1 2027, effectively taking Nvidia from a global chip giant powering the world’s computing to one that is more-or-less wholly dependent on American spending and policy.

The company recognizing this transition by announcing in n Q1 2027 that it has begun to change its presentation of revenue by market platform. Henceforth, this breakdown will be:

• Hyperscale: one of two “Data Center” sub-segments, it includes revenue generated from public cloud providers such as Microsoft Azure, Google Cloud and Oracle Cloud as well as Consumer Internet Companies such as Meta.

• ACIE (AI Clouds, Industrial, and Enterprise): the second “Data Center” sub-segment, it includes purpose-built, highly diverse AI infrastructure deployed outside of traditional hyperscale environments such as specialized “AI Clouds” that focus entirely on GPU-accelerated workloads, large-scale physical applications such as factory automations and telecom AI deployments, corporations building proprietary "AI Factories", as well as national governments and state-backed entities purchasing computing power to build localized domestic AI capacity.

• Edge Computing: this segment provides data processing devices that power decentralized, agentic AI and physical AI. It explicitly includes AI PCs and Workstations, Gaming, Automotive, and embedded systems.

Nvidia indicated in the Q2 27 release that the ACIE segment surged 138% YoY in Q2 2027 and grew 114% in H1 2027 relative to H1 2026 to deliver $72.5 billion throughout H1 2027. The ACIE segment merits an examination in light of an important report regarding an upcoming acquisition by Nvidia.

The Unmodelled Consequences of Alignment

Various estimates contend that around $30-40 billion of sovereign AI projects are underway. While this number is around the same as the total amount of inventory Nvidia alone has, it is an important foundational investment made by nation-states to exert sovereignty in AI development and safeguarding homegrown industries from excessive dependence on foreign entities.

Within major sovereign projects presently underway, the Middle East (mostly Saudi Arabia and UAE) accounts for about 30% of project value while the EU and Japan account for 23% and 18% respectively. All told, these three clusters account for a little over 70% of estimated sovereign projects by value. In the pre-AI age, Nvidia could have been deemed a “neutral entity” that could have found favour within such types of projects. However, he US government’s “Pax Silica” policy – which emphasizes the primacy of the American “tech stack” and has Nvidia squarely within – essentially turns Nvidia into an instrument of American policy rather than a globally-competitive entity. Alternative compute hardware suppliers could be expected to rise wherein they would be increasingly tied to proximity and alignment to blocs rather than the full and true depth of the global compute market.

In an interesting move, reports suggest that Nvidia has moved to acquire HuggingFace, the central repository and collaboration hub for open-source AI models, weights, and datasets, in a calculated move to control AI’s software development layer. While Hugging Face had historically remained hardware-agnostic, its post-acquisition timeline could include default optimizations, model runtimes, and backends that natively embed Nvidia’s CUDA, TensorRT-LLM, and NIM microservices – thus widening the friction for developers attempting to compile onto competing hardware. This runs counter to several objectives of sovereign compute initiatives and should intensify questions from all investors in sovereign AI – two of whom (the EU and Japan) have fairly mature software industries and global collaboration networks.

This move also acknowledges a critical development in the AI buildout currently: custom silicon – as developed by Amazon and Google – have been increasingly emplaced with hyperscaler clusters and impacting Nvidia’s potential market growth outlook. Nvidia’s acquisition of HuggingFace brings to the forefront the tension the company faces without actually resolving it: other open-source repositories are bound to rise and will (possibly) be shielded from behind sovereign walls. Whether Nvidia would be able to truly access the full breadth and depth of the global AI market would be an important question in the quarters and years to come.

Market Reaction and Impact

During the earnings call, Nvidia flagged tight memory supply and rising component costs – which led to an immediate uptick in Asian stocks such as SK Hynix, Samsung Electronics, TDK and Kioxia. The biggest jump was in the price of Largan Precision due to the news that the Vera Rubin platform is entering full production. The Vera Rubin platform includes Spectrum-6 networking with co-packaged optics, with the latter being Largan’s unique niche within Taiwan’s semiconductor industry. The enthusiasm, however, didn’t transfer substantially to Nvidia’s fab partner TSMC – which is a telling sign.

The immediate reaction from the market in the after-hours session to Nvidia’s release was mildly bearish and driven by the fact that margin erosion is inevitable if memory and component costs stay elevated. The acquisition and attempts at consolidating at the software layer should be creating signals for the market that the challenge from custom silicon is real, resilient and being felt by the company. As a result, Nvidia would have less room to manoeuvre and continue growth in light of shrinking product mix and geographic diversification as well as the consequences of American policies and the fragmentation of the global order. Alternatives to Nvidia will rise and its decision to acquire HuggingFace serves to make the case for incubating and developing alternatives much stronger.

In terms of stock valuation, Nvidia announced that it returned nearly $26 billion to shareholders via share buybacks and dividends in Q2 alone while maintaining an active $99 billion repurchase authorization. While this will prove to be some sort of support to the company’s stock valuation, the seemingly momentous change in long-term outlook should intuitively be hard to ignore.

By centering on US datacenter/hyperscaler spends at a moment of intense opposition to datacenters being felt across the US and the move to consolidate the “tech stack” under the aegis of American policy, Nvidia’s vision for global relevance is likely a matter that will be questioned and addressed in the years to come – not by the market per se but by those who will drive policies around the world. By virtue of its alignment, Nvidia might not have much by way of assurances for the latter.

Investors with access to European might want to consider the +3x Long NVIDIA ETP ( $LS 3X NVIDIA(NVD3.UK)$ ) and the -3x Short NVIDIA ETP ( $LS -3X SHORT NVIDIA (NVDA) ETP(NV3S.UK)$ ) during bullish and bearish trends in NVDA’s price. To potentially capitalize on major tech stocks seemingly driving the market currently, the 5x Long Magnificent 7 ETP (MAG7) and the -3x Short Magnificent 7 ETP ( $-3X SHORT MAGNIFICENT 7 ETP(MAGS.UK)$) are at hand.

Furthermore, the NVIDIA Options ETP (LSE ticker: NVDI) seeks to generate monthly income by buying NVIDIA shares and selling put options on them. Also available is the Magnificent 7 Options ETP (MAGO), which invests in each Magnificent 7 constituent’s respective Options ETP in an equally-weighted manner.