Nvidia 2Q FY27: The Party Continues

The party continues

As AI run race continue, AI models are no longer tools appeared on headlines, but tools we are all using, and the sustainability of AI computing demand is always an open-ended questions every investors have been questioning about.

Unsurprisingly, on 26 August 2026, Nvidia once again delivered a double beat on both top and bottom-line growth, defending its place in the AI chips space, while guided an upbeat guidance on upcoming sales growth, strengthening the bull says. Surprisingly, market reacted this stellar earnings results positively, in contrast to the previous 4 consecutive quarter results where market sold off the shares despite the consistent earnings results beat.

In short, 1Q FY27 deepened the moat, with Vera Rubin production shipments appear ahead of schedule, pushing the digestion risks we flagged further out, while margin remain stable at 75%, despite guided a lower margin due to elevated memory cost. To reiterate, as the world races to build its thinking machines, Nvidia is increasingly monetising adjacent layers of the AI infrastructure stack beyond accelerators alone.

As such, we remain positive on Nvidia.

Figure 1: Nvidia’s share prices.

Another clean quarter

Nvidia delivered another beat on both the top and bottom lines, with revenue more than doubling, rising 106% y/y and 18% q/q to USD96.2bn. Data centre remained the main growth engine, with revenue increasing 117% y/y and 18% q/q to USD89.0bn.

Within data centre, Hyperscale revenue grew 13% q/q to USD48.7bn, supported by sustained Blackwell demand. Meanwhile, AI Clouds, Industrial and Enterprise, or ACIE, expanded faster at 25% q/q to USD40.3bn. This shows that Nvidia’s growth is broadening beyond direct purchases by several large hyperscalers, despite part of the ACIE demand still ultimately supports hyperscaler and frontier-lab workloads through AI cloud providers.

Gross margin met guidance and remained stable q/q at 75.0%, while improving 2.5% y/y, reflecting the stronger Blackwell Ultra mix offsetting higher system and component costs during the quarter.

Cash conversion, however, weakened sharply. 2Q FY27 free cash flow increased 59% y/y but fell -56% q/q, mainly due to higher working capital and cash taxes. On top of that, inventory increased to USD31.6bn ahead of the Rubin ramp, while days sales outstanding rose from 45 to 60 days, reflecting extended payment terms on large, multi-quarter agreements with certain customers. Nevertheless, we do not see the inventory increase as a sign of demand slowing, as it coincides with the Rubin production ramp and remains supported by a significantly stronger revenue outlook.

Figure 2: Nvidia’s 2Q FY27 key financial performances.

Source: Nvidia, Claude compilations.

Figure 3: Data centre growth remains strong.

Vera Rubin platform has entered full production

The key product development in the earnings call is that Vera Rubin platform is now entering full production. Production shipments commenced in early August, while Nvidia has already received purchase orders from every major hyperscaler, AI cloud and system OEM. On top of that, management expects Rubin to account for approximately 20% of Data center revenue in 3Q FY27 and become the fastest product ramp in Nvidia’s history, materially reducing the risk of a revenue gap as Blackwell begins to wind down.

More importantly, Rubin expands Nvidia’s revenue opportunity beyond the GPU. The Vera CPU has entered full production, with shipments already underway to lead customers, and the company continues to see around USD20bn of demand for server CPUs and expects CPU revenue to more than double in FY28. Together with the Rubin GPU, networking and other system components, we believe the more integrated platform could increase Nvidia’s revenue contribution per gigawatt from around USD25bn under Blackwell to USD40bn under Vera Rubin.

With Blackwell demand remaining firm as Rubin begins to ramp, customers seem to not delay deployments ahead of the new architecture. This points to an overlapping product cycle and reduces the risk of a near-term revenue gap between Blackwell and Rubin. Hence, despite digestion risk remains, the broad order coverage and rising revenue content per deployment is likely to extend Nvidia’s earnings visibility into FY28, pushing the slowdown in the current cycle further.

Growth is broadening beyond major US hyperscalers

Notably, while hyperscalers historically remain Nvidia’s largest AI data centre customers, there are increasing rapid expansion from enterprises, neoclouds and sovereign AI projects, supported by large deployment plans from customers such as SpaceX, Oracle and regional AI cloud providers. This gives Nvidia more routes to market and reduces its direct reliance on the spending decisions of a small number of US cloud providers.

Meanwhile, Nvidia is also earning more from every data centre project, with revenue contribution per gigawatt has risen from around USD18bn under Hopper to USD25bn under Blackwell and is expected to reach USD40bn under Vera Rubin, reflecting Nvidia’s continue expansion from GPUs into CPUs, networking, switching and specialised inference processors. In other words, growth is coming not only from more data centre capacity, but also from higher Nvidia content within each deployment.

Figure 4: ACIE growth outpacing hyperscalers.

Memory scarcity is the main bottleneck to AI data centre buildout

While Nvidia’s FY28 revenue outlook was significantly stronger than expected, with management guiding for growth of approximately 70%, customer forecasts point towards roughly 100% growth, where the lower revenue guidance was largely due to the constraint in memory and broader supply chain capacity.

Notably, the company’s commitments increased 134% q/q from USD119bn to USD279bn, primarily for memory procurement, yet, the additional supply is coming at a higher cost, with management expecting gross margin to decline from 75% in 2Q to 74% in 3Q, before falling to around 71–72% in 4Q. Meanwhile, although the company recently raised prices by 15%, we believe the lower margin guidance suggests that Nvidia will absorb part of the cost increase during the Rubin ramp rather than passing it through immediately.

Despite lower margin, at 71-72% margin, in our view, is still at a very robust level, especially amid this rising memory cost, slightly below the past 3 years average quarterly margin, at 74%, and is still does not outweigh the scale of the broader revenue upgrade. We therefore upwardly adjust revenue growth to drive higher absolute gross profit and earnings than we currently previously modelled.

Figure 5: Nvidia gross margin. 

Figure 6: Management guidance ahead.

Source: Nvidia, Claude compilations.

Vendor financing risk becomes more visible

Since late last year, market discussion on the “AI closed-look economy” or so called AI circular financing has never been faded, where to some extent, we believe this is part of the reasons why Nvidia multiples has been trading at the lower level.

That said, it is encouraging to see that Nvidia provided substantially more disclosure on its supply commitments, credit exposure, leases and customer-support arrangements during the quarter, to some extent, could ease investor’s concerns on vendor financing and the demand profile beyond the top five cloud service providers.

On the USD500bn third-party financing platform targeted through Nvidia’s partnerships with infrastructure capital providers, we believe it could extend the AI investment cycle as Nvidia systems are widely adopted, transferable across operators and supported by CUDA, giving lenders greater confidence in their utilisation and residual value. In another way, we believe even if one operator fails, the Nvidia systems could be transferred to another operator, helping preserve their collateral value.

However, it is undeniable that Nvidia’s growing on- and off-balance-sheet commitments remain a concern, particularly as its total exposure is still difficult to quantify. If a major participant such as OpenAI were to face financial difficulties, Nvidia would be among the most directly exposed suppliers and capital providers. The risk would also extend beyond the semiconductor sector, as the latest financing structures involve a broader group of banks, private credit funds and infrastructure investors.

While this remains a tail risk rather than our base case, its potential impact has increased as Nvidia becomes more involved in financing the ecosystem supporting its own demand.

Figure 7: Nvidia newly disclosed commitments.

Reiterate our BUY rating.

We maintain our BUY rating. Nvidia’s FY28 revenue outlook, continued Blackwell strength and early Rubin production collectively extend the earnings runway beyond our current assumptions. Rubin has also reduced the risk of a near-term transition gap, while supply constraints leave room for additional upside if more memory and manufacturing capacity becomes available.

Our primary methodology values the equity on FY2028E EPS multiplied by a fair price/earnings multiple. Based on our fair P/E of 24x, we derive an upside of 79%, with a target price of USD375. Thus, we maintain our BUY rating.

Table 1: Valuation summary.

	FY26A	FY27E	FY28E	FY29E
Revenue (in USD millions)	215,938.00	405,921.20	588,585.74	718,074.60
Growth %, y/y	65.50%	87.98%	45.00%	22.00%
Adjusted EPS	4.36	9.06	12.53	15.63
Growth %, y/y	48.64%	107.77%	38.32%	24.72%
P/E at current price	48.17	23.18	16.76	13.44
Fair P/E (x)				24
Upside Potential				78.6%
Target Price				375
Source: Bloomberg Finance L.P., iFAST compilations. Data as of 26 August 2026.

Investment Risks:

The principal downside risks are:

1.      ROI uncertainty and reduced hyperscaler’s capital expenditure could slows revenue and compresses the multiple simultaneously.

2.      NVDA continues to invest in on/off balance sheet commitments.

3.      Accelerating custom-silicon adoption eroding accelerator share.

Declaration:

This research report was prepared with the assistance of artificial intelligence (AI) tools. iFAST Financial Pte Ltd does not rely exclusively on AI for content generation; the content of this report – including all investment theses, ratings, price targets and conclusions – has been independently reviewed and verified by the research analyst(s) to ensure accuracy and professional integrity.