Nvidia's hyperscaler dependence is again emerging as a key variable in its results for the quarter.

On Aug. 25, CNBC reported that Wall Street broadly agrees Nvidia is the biggest beneficiary of the artificial intelligence boom. But it sees the company’s structure, with revenue concentrated among a small number of large customers such as Amazon, Google and Microsoft, as a test for long-term growth. Meta and SpaceX are also building their own AI models, buying graphics processing units (GPUs) in large volumes and starting to rent excess supply to companies that need it.

Nvidia has been disclosing its data center customer base split into hyperscalers and ACIE since its earnings report in May. ACIE is a category combining AI cloud, industry and enterprise segments. Nvidia did not name specific companies that fall into the hyperscaler group, but Nvidia Chief Executive Jensen Huang (젠슨 황) defined the group relatively clearly.

In the May earnings report, Huang said, "The easiest path to market is, of course, hyperscalers, because there are only five or six." He added, "The rest of the industry is 250,000 companies worldwide."

At the time, first-quarter revenue was almost similar between the two segments. Hyperscaler revenue was $37.9 billion and ACIE revenue was about $37.5 billion. But ACIE grew 31 percent quarter-on-quarter, outpacing the hyperscalers' 12 percent. That amounted to Nvidia itself revealing the task of maintaining a revenue structure centered on a small number of large customers while expanding broader enterprise demand.

The market has already begun to question that. Nvidia shares fell 2.9 percent last Monday, extending weakness for seven straight sessions. It was the longest losing streak since 2022, with a cumulative decline of 7.5 percent over the period. Investors want Nvidia to reduce its dependence on a small customer group while also worrying whether those customers have room to increase spending further. Amazon and Alphabet swung to negative free cash flow in the second quarter, and Meta's cash generation fell by more than 90 percent from a year earlier. SpaceX and Tesla, led by Elon Musk, also recorded negative free cash flow as they expanded AI investment.

Gene Munster (진 먼스터), managing partner at Deepwater Asset Management, said in an interview, "Behind this issue is investors' concern about how sustainable Nvidia's rise is and that hyperscalers will no longer be able to significantly increase spending." He added, "They want to see whether other customer groups are starting to move in earnest."

Nvidia rewrote and published results for the latest nine quarters on its website in May. The data showed hyperscalers accounted for about 55 percent of data center revenue over the latest year. In the most recent quarter, hyperscaler revenue rose 115 percent year-on-year, outpacing ACIE's 74 percent growth.

Analysts expect that trend to reverse going forward. Based on StreetAccount tallies, ACIE revenue is expected in second-quarter results at $43.0 billion, up 149 percent over the past year, while hyperscaler revenue is seen at $43.6 billion, up 83 percent. SpaceX is cited as one of the key growth drivers for the hyperscaler business. Munster said SpaceX is looking to rapidly build Nvidia-based data centers over the next year.

According to LSEG tallies, analysts expect Nvidia's total revenue to nearly double from a year earlier to $92.2 billion. The data center segment is taking a larger share of revenue. StreetAccount estimates put data center revenue at $86.3 billion, expected to account for 94 percent of the total. That is up from 92 percent in the first quarter. LSEG said, on an annual basis, revenue is expected to rise 83 percent this year to $396.0 billion, then growth is forecast to slow to 44 percent next year.

To diversify its customer base, Nvidia is moving proactively rather than waiting for new customers to arrive. Recognising that few companies have the cash or financial capacity to buy rack-level systems, it is working with Wall Street on efforts to make GPUs an asset class.

Earlier this month, Nvidia unveiled a financial programme with six major financial firms that could raise up to $500.0 billion from investors who view semiconductors as an investable asset like real estate. Huang said the structure would allow more companies to finance GPU purchases at lower interest rates as semiconductors generate returns. Nvidia said at the time of the announcement it had signed memorandums of understanding with the investment firms, but many details have yet to be disclosed.

Another metric to watch in this earnings report is revenue from the Vera Rubin system, whose shipments have only recently begun to expand. Earlier this year at its GTC conference, Huang said the company expects cumulative revenue of $1.0 trillion through 2027 via the current-generation Blackwell and Vera Rubin.

If Vera Rubin performs as a bigger success than expected, concerns about dependence on a small customer group could ease somewhat. Analysts at KeyBanc, which has a buy rating, said over the weekend that expanding Rubin GPU shipments would be a key driver for earnings upgrades, and they expect strong results and guidance.