
NVIDIA's Business and Position
NVIDIA (NVDA) is the world's largest publicly traded company. It is known for graphics processing units (GPUs) but has grown into a full computing platform built around AI and accelerated computing.
Earlier this year NVDA changed its reporting into two segments: data center and edge computing. Data center is by far the largest, bringing in $89 billion last quarter, over 90% of total revenue. It covers GPUs, CPUs, networking gear, and software used to build and run data centers, which NVDA calls "AI factories." Edge computing brought in $7.2 billion and covers products used outside traditional data centers: PCs, workstations, gaming, automotive, robotics, and other physical AI uses.
Competitors
AMD (AMD) is NVDA's most direct rival in high-performance data center GPUs. Broadcom (AVGO) has become an important competitor through custom AI accelerators built for large cloud and AI customers. Some of NVDA's biggest customers are also turning into competitors: Alphabet (GOOGL), Amazon (AMZN), Meta (META), and OpenAI are all building their own chips. The point of these custom chips is to cut costs and reduce reliance on NVDA hardware for certain AI workloads, not to fully replace NVDA.
NVDA's main edge is that it does not just sell chips. It has built a full computing ecosystem combining GPUs, CPUs, networking, and software. At its center sits CUDA, NVDA's software platform that developers have used for years to build apps around its hardware. That installed software base is a strong advantage because customers can use the same NVDA architecture across cloud providers, enterprise data centers, and different hardware generations. NVDA also moves fast through product cycles: Blackwell drives the current wave of AI spending, and the next-generation Vera Rubin platform is already ramping into full production.
The Earnings
- Q2 revenue: $96.2 billion, up 106% from a year ago, well above the roughly $92 billion expected.
- Non-GAAP earnings per share: $2.22, above the $2.09 expected.
- Gross margin held at 75% despite rapid growth.
- Data center revenue: $89 billion, up 117% year over year.
- Edge computing revenue: $7.2 billion, up 27%.
The biggest news was guidance. NVDA guided Q3 revenue to $108 billion, plus or minus 2%, above expectations. The CFO said NVDA now expects revenue to grow about 70% in fiscal 2028, far above the roughly 45% Wall Street had expected. CEO Jensen Huang said real demand is even stronger, but NVDA is limited by how much its supply chain can produce. That longer forecast eased fears that AI infrastructure spending could be near a peak.
The Positives
First, AI demand is broadening. The early build-out was dominated by a few hyperscalers and large AI labs. Demand is now spreading across frontier AI labs, enterprise customers, sovereign AI projects, industrial companies, and physical AI uses. NVDA's AI cloud, industrial, and enterprise customers generated more than $40 billion in sales during the quarter, up 138% year over year. This spread reduces dependence on a handful of big cloud customers.
Second, NVDA is still supply constrained. That is an operational problem, but it also shows demand keeps outrunning what the company can deliver.
Third, Vera Rubin is already entering production, giving NVDA another product cycle after Blackwell. The company is also expanding beyond GPUs into networking, CPUs, software, and complete data center systems, letting it capture a larger share of each AI project.
Fourth, the business generates huge profits and cash flow. NVDA returned about $26 billion through buybacks and dividends and still has about $99 billion left under its share repurchase authorization.
The Risks
Margins. The 75% gross margin is expected to slip to 74% next quarter and could fall further as rising memory prices push up the cost of building NVDA systems.
China. Q3 guidance assumes zero data center compute revenue from China. Export restrictions have sharply cut NVDA's access to what was once an important market, and domestic Chinese competitors keep developing their own AI hardware.
Custom chips. Alphabet (GOOGL), Amazon (AMZN), Meta (META), OpenAI, and other large customers have strong financial reasons to design processors more efficient for their own workloads.
Circular financing. Critics say NVDA can help finance customers who then use that money to buy NVDA equipment, funding the very build-out that drives demand for its products.
The Chart
After earnings NVDA has pulled ahead of the S&P 500, up about 26% versus the index's 19%, though still behind the tech sector overall. That gap reflects NVDA's earlier strong multi-year rally, with recent price moving more sideways in a consolidation phase.
Against rivals in the CPU, GPU, and specialized chip space, Intel (INTC) is the dominant name on a yearly basis. Intel owns its own domestic US foundry. NVDA and many others on the list are "fabless" companies: they only design chips and send the designs to another firm, such as Taiwan Semiconductor (TSM), to be built.
Key price levels:
- All-time high: 236.54.
- A high near 227 around August 13-14, followed by a collapse into earnings.
- Before earnings, price formed a small, narrow, downward-sloping channel. Earnings broke price above the gap near 225 that formed during the decline. That old gap now forms a support zone from about 225 up to the old highs near 227.
- Upside level to watch: 233, a notable intraday high from June 2.
- Downside levels: 212, a repeated ceiling during a range-bound stretch, and 207, the extreme low of the recent decline.
Moving averages are not much help after a sharp move because they lag, being based on past prices. The most useful now is the 5-day weekly exponential moving average, the shortest term at just above 218, which should act as short-term support as it catches up.
RSI, the momentum measure, broke above its downward trendline and at the same time crossed above the 50 midline that splits bullish from bearish momentum. That is a doubly bullish read after earnings.
The volume profile shows price pushing past heavily traded areas. The last small node sits around 221-222, and price is now above it, moving into thinly traded territory that could bring fast price moves.
The options market's expected move points to possible new all-time highs by September. The September 18 monthly expiration implies an expected move of about 7.2%, with an upper boundary at 242.30, above the old highs.
Bottom Line
NVDA did something increasingly hard: it beat already enormous expectations. Demand for its AI chips is no longer the question. The bigger questions are how long this level of AI infrastructure spending can last, whether NVDA can keep its extraordinary margins as costs rise, and how much market share custom chips can eventually take. NVDA sits at the center of the global AI build-out. The central question for investors is whether that build-out can keep growing fast enough and create enough economic value for NVDA's customers to support the company's own growth for years to come.


