Nvidia

$96.22B in Q2 2026 revenue, up 106% year over year, underscored Nvidia’s shift from AI-chip supplier to a broader AI infrastructure platform under export pressure.

Who they are

Nvidia appears in the coverage as the central supplier of AI computing hardware and infrastructure, with its GPUs and flagship AI chips linking cloud companies, model developers and data-center builders. Stories also place it beyond discrete accelerators: it is extending into PC chips with RTX Spark, interconnect technology through NVLink Fusion, and AI-model ecosystems through Nemotron.

The recent arc

Coverage accelerated through late 2025 and into 2026, reaching its recent high in 2026Q3 as the story moved from demand for AI chips to the scale and reach of Nvidia’s platform. Huang’s projection of more than $1 trillion in flagship AI-chip sales through 2027, followed by Q2 revenue of $96.22 billion and $89 billion in data-center revenue, made financial execution the anchor of the period.

The latest stories show Nvidia being pulled into adjacent layers of AI deployment. It launched the Arm-based RTX Spark consumer-chip family with TSMC and MediaTek; Apple is reportedly considering AI servers using its M8 Ultra chips with Nvidia’s NVLink Fusion; Salesforce’s Koa is built on Nvidia’s Nemotron architecture; and Google, Nvidia and Emerald AI launched an energy-management alliance for responsive data centers. Meanwhile, US policy remained a defining constraint, from the conditional opening for H200 shipments to approved Chinese customers to reported revenue-sharing terms attached to China export licenses.

The tension

The core tension is Nvidia’s effort to preserve AI-infrastructure leadership while customers, rivals and governments reshape the market around it. AMD and GPU startups are competing for AI-chip demand, Apple is exploring more of its own server silicon, and Nvidia’s relationship with Intel is notably mixed: it bought roughly $5 billion of Intel shares but reportedly did not proceed with Intel’s 18A process. Its non-exclusive Groq inference-technology license and recruitment of Groq executives similarly show it acquiring capabilities and talent while limiting a potential source of competition.

Why it matters

If this trajectory holds, Nvidia’s influence could increasingly depend on control of the connections among chips, models, servers and power management rather than GPU sales alone. That would make partnerships with companies such as Apple, Google and Salesforce strategically important, but it also exposes the company to a wider set of risks: China export rules, customers’ in-house silicon efforts, and competitors targeting inference or alternative AI architectures.

Related: China · Microsoft · AI · Intel · AMD · Jensen Huang