What happened
Nvidia reported revenue of 96.2 billion dollars for the second quarter of its fiscal 2027 year, up 106 percent from a year earlier and 18 percent from the previous quarter. The results, published on 26 August, beat analyst expectations and sent the shares up 8.7 percent.
Data centre revenue, the segment that sells the chips powering artificial intelligence systems, reached 89.0 billion dollars. That is 117 percent higher than a year ago and accounts for more than nine tenths of the company total. Growth was driven by the ramp up of Blackwell Ultra, the current generation of Nvidia AI infrastructure.
Margins held up. Gross margin, the share of revenue left after the direct cost of making the product, came in at 75.0 percent on both a reported and adjusted basis. Earnings per diluted share were 2.46 dollars reported and 2.22 dollars adjusted.
The guidance was the bigger story. Nvidia told investors to expect roughly 108 billion dollars of sales in the third quarter, plus or minus 2 percent. Chief executive Jensen Huang went further, forecasting revenue growth of about 70 percent for fiscal 2028, far above what analysts had modelled.
Why it matters
Nvidia is now large enough that its results function as an economic indicator rather than a company update. A single quarter of 96 billion dollars is more revenue than most FTSE 100 companies generate in a year.
The read across runs through the whole market. Nvidia sells the picks and shovels of the AI build out, so its order book is the clearest available evidence of how much Microsoft, Amazon, Alphabet, Meta and a long tail of specialist providers are actually spending on computing capacity. A 117 percent jump in data centre sales says that spending has not slowed.
It matters for energy and infrastructure too. Williams Companies, a US gas pipeline operator, rose 5.6 percent this week on demand from AI data centres. Those facilities consume electricity at industrial scale, which is reshaping power markets, grid investment and gas demand in ways that have nothing to do with semiconductors.
There is a concentration risk on the other side. When one company drives so much of the index, a disappointment becomes a market wide event rather than a stock specific one. Analysts currently project 30.0 percent earnings growth for the S&P 500 in calendar 2026 on 11.5 percent revenue growth, and a large slice of that sits with a handful of technology names.
Explained simply
Every company racing to build artificial intelligence is digging in the same gold rush, and Nvidia sells the only shovels that work. It does not matter which prospector strikes it rich.
An AI model is trained by performing an enormous number of simple calculations at the same time. Ordinary computer processors handle tasks one after another very quickly. Nvidia graphics processors handle thousands simultaneously, which is why they became the standard tool for this work almost by accident, having originally been designed for video games.
A data centre is simply a warehouse full of these chips, wired together, cooled and connected to a large electricity supply. When Nvidia reports data centre revenue, it is reporting how many warehouses the world is filling.
Blackwell Ultra is the current chip generation. Each generation delivers more calculations per unit of electricity, which is why customers upgrade rather than simply buying more of the old model. The ramp Nvidia refers to is the process of shifting production and shipments to that newer design.
Gross margin at 75 percent tells you how little competition there is. For every 100 dollars of chips sold, 75 dollars remains before research, sales and administration costs. Manufacturers in genuinely competitive markets typically operate on a fraction of that.
What it means for you
If you hold a global tracker fund, a US index fund or a workplace pension with a default global equity option, you own Nvidia whether you chose to or not. It is one of the largest constituents of the S&P 500 and of the MSCI World index, which means it can easily represent 5 percent or more of a supposedly diversified fund.
That is worth checking rather than assuming. Look at the top ten holdings of your pension default fund. If the first five names are all US technology companies, the fund is far more concentrated than the word global suggests, and its performance over the next two years will be decided largely by whether AI spending holds up.
UK investors have less direct exposure. The FTSE 100 has almost no large scale semiconductor presence, which is why it has lagged US indices through this cycle. Splitting a portfolio between a FTSE tracker and a global fund delivers diversification of a real kind here, not just a different label.
Anyone tempted to buy the individual share should note what is already in the price. Growth of 106 percent is extraordinary, but the guidance implies the market expects extraordinary growth to continue. Meeting expectations is not enough at these levels; the shares need surprises.
The bigger picture
The AI capital expenditure cycle is now the dominant force in global equity markets, and it is a capital spending story more than a software story. Five of eleven S&P 500 sectors are reporting double digit earnings growth, led by energy, information technology and communication services, and the common factor in all three is infrastructure being built at speed.
The unanswered question is return on investment. Enormous sums are being spent on capacity in the expectation that AI products will eventually generate matching revenue. Nvidia gets paid either way, which is precisely why its numbers are strong and why they are not, on their own, proof that the wider bet is working.
What to watch next: capital expenditure guidance from the large cloud providers at their next results, electricity price and grid connection data in the US, and whether Nvidia gross margin starts to slip as rivals ship competing hardware.
