Nvidia Earnings Forecast Looks Cheap on Piper's Math, but Execution Is the Real Test

Nvidia has received another bullish rating, despite an Nvidia earnings forecast that demands extraordinary growth through fiscal 2030. Piper Sandler initiated coverage with an Overweight rating and a target suggesting roughly one-third upside from the reference share price.

The valuation initially looks unusually modest for the leading supplier of AI computing systems. Piper estimates that Nvidia trades near 14 times its fiscal 2028 earnings and carries a price-to-earnings-growth ratio near 0.3.

However, that apparent discount depends on earnings reaching about 30 per share by fiscal 2030. Piper expects revenue to compound at 47% annually through that period, with earnings expanding at a similar rate.

That is the central conflict. Nvidia looks inexpensive only after investors accept a growth curve that would turn today's exceptional demand into several more years of exceptional execution.

Piper's Nvidia Earnings Forecast Resets the Valuation Debate

Piper's rating reframes Nvidia as a growth stock with a surprisingly low forward multiple, rather than an expensive symbol of AI enthusiasm.

Piper Sandler initiated coverage in early September with an Overweight rating. Analyst David O'Connor based the recommendation on Nvidia's position in AI compute, its annual product cadence, and continuing supply constraints.

The firm's Nvidia coverage estimates a 47% compound annual revenue growth rate through fiscal 2030. It expects earnings to grow at a comparable pace and reach approximately 30 per share.

Piper also estimates that Nvidia controls about 80% of the AI compute market and roughly half of unit shipments. Those figures describe a market where Nvidia sells more valuable systems than its unit share alone suggests.

The analyst's valuation uses approximately 14 times calendar 2028 earnings. That produces a price-to-earnings-growth ratio, or PEG ratio, of about 0.3.

A PEG ratio compares a company's earnings multiple with its expected growth rate. Investors often use it to ask whether a high valuation remains reasonable relative to projected expansion.

On this framework, Nvidia appears inexpensive. A company growing earnings near 47% annually would usually command more than 14 times forward earnings, assuming that growth remains credible.

The arithmetic is internally consistent. The difficulty lies in the forecast period and the scale of the required results.

Fiscal 2030 is distant enough for product cycles, customer budgets, export policies, and competitive positions to change. Small revisions to revenue growth or margins can produce large differences in the eventual earnings figure.

The 14-times multiple also looks further forward than a standard next-year valuation. It discounts earnings that Nvidia has not yet generated across several demanding product transitions.

The Piper valuation case therefore does more than identify a low multiple. It asks investors to decide how much confidence distant estimates deserve.

Piper's argument starts from visible demand. Agentic AI systems, which can plan and execute multistep tasks, require substantial computing capacity during both training and operation.

The firm expects AI compute supply to remain constrained for another two to three years. It also points to higher rental rates for existing GPUs as evidence that available capacity remains valuable.

Those signals support a strong near-term outlook. They do not automatically validate the same growth rate through fiscal 2030.

A constrained market can attract more supply, alternative architectures, and customer-designed chips. It can also encourage buyers to improve software efficiency or delay less productive deployments.

Piper's initiation changed the framing around NVDA. The debate is no longer simply whether AI demand remains strong this quarter.

The harder question is whether Nvidia can convert that demand into enough durable profit to make a distant multiple relevant today.

Nvidia's Latest Results Make the Bull Case Plausible

Nvidia's current financial performance gives Piper a credible starting point, but it also raises the base that every future comparison must exceed.

Nvidia reported fiscal second-quarter 2027 revenue of 96.2 billion for the period ending July 26, 2026. Revenue increased 18% sequentially and 106% from the prior year.

Data Center revenue reached 89.0 billion, up 18% sequentially and 117% year over year. That business generated more than nine-tenths of total quarterly revenue.

The company's quarterly results also showed a 75.0% GAAP gross margin. GAAP diluted earnings per share reached 2.46, while non-GAAP diluted earnings per share reached 2.22.

GAAP figures follow standard accounting rules. Non-GAAP results remove selected items that management believes can obscure operating comparisons.

Both measures confirm that Nvidia is converting revenue into unusually large profits. GAAP operating income reached 63.7 billion, more than double the result from one year earlier.

These numbers explain why a distant earnings estimate can appear achievable. Nvidia has already demonstrated that demand can grow faster than its enormous revenue base.

Management's next-quarter outlook extended that momentum. Nvidia expects fiscal third-quarter revenue of 108.0 billion, within a range of 2% above or below that figure.

The company expects GAAP and non-GAAP gross margins near 74.0%, with a possible variation of 50 basis points. One basis point equals one-hundredth of a percentage point.

Importantly, the outlook assumes no Data Center compute revenue from China. Nvidia is therefore projecting another record quarter without relying on a major market that once offered meaningful growth potential.

Nvidia also told investors that it expects approximately 70% revenue growth in fiscal 2028. Management described that outlook as supply constrained, meaning available production remains below expected customer demand.

During the earnings call, Chief Financial Officer Colette Kress said cloud industry backlog exceeded 2 trillion. She also cited rapidly rising planned capital spending among the largest hyperscalers.

Hyperscalers are cloud companies operating massive computing networks. Their spending matters because they are among the largest buyers of Nvidia accelerators, networking equipment, and complete rack-scale systems.

That customer demand extends beyond model training. AI laboratories, cloud providers, enterprises, governments, and emerging AI companies increasingly need capacity for inference, which runs trained models for users.

Nvidia's product strategy also supports the forecast. The company now sells integrated systems combining accelerators, networking, interconnects, CPUs, and software.

This approach raises the value attached to each deployment. It also makes comparisons based only on GPU unit shipments less informative.

The Vera Rubin platform has entered production, according to Nvidia. Partners operating early systems include Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, CoreWeave, and Nebius.

An annual architecture cadence gives Nvidia repeated opportunities to improve performance and encourage customers to upgrade. It also creates a demanding execution schedule across manufacturing, networking, cooling, and software.

The current numbers validate Piper's claim that Nvidia leads the market. They also show why the Nvidia earnings forecast faces a mathematical challenge.

When revenue already exceeds 96 billion in one quarter, maintaining a 47% annual growth rate requires enormous additional spending. Each successive percentage point represents more absolute revenue than before.

The bull case is credible because Nvidia has repeatedly cleared high expectations. It remains demanding because past acceleration has created a much larger base.

The Real Opponent Is Piper's Forecast, Not AMD

Nvidia does not need to eliminate every competitor, but it must outperform the assumptions already embedded in Piper's valuation.

AMD is the most visible merchant competitor in data center accelerators. It offers Instinct GPUs, EPYC server processors, and an expanding software environment for AI workloads.

AMD reported second-quarter 2026 revenue of 11.5 billion. Its Data Center business more than doubled from the prior year, according to AMD's quarterly results.

That growth matters because customers want alternatives. Cloud providers benefit from negotiating leverage, supply diversity, and hardware designed for different workloads.

Google, Amazon, Microsoft, and other large customers also develop custom accelerators. These chips can reduce dependence on merchant GPUs for carefully defined internal tasks.

Even so, competition does not need to take Nvidia's market leadership away to affect Piper's forecast. It only needs to slow pricing, weaken margins, or capture enough incremental demand.

A forecast based on 47% annual growth leaves less room for gradual share loss than a stable-growth model. Nvidia must preserve both market access and the economic value of its systems.

The company's defense extends beyond accelerator performance. CUDA software, networking, development libraries, and optimized models make Nvidia hardware easier to deploy across many applications.

Customers also buy complete systems rather than isolated components. This reduces integration work, but it can increase commitment to Nvidia's architecture.

That installed base creates switching costs. Developers have already written software around Nvidia tools, while operators have trained staff around its systems.

However, switching costs are not permanent protection. Open software layers can make workloads more portable, and large cloud companies have strong incentives to support their own silicon.

Efficiency improvements create another form of competition. A model that needs fewer processors for training or inference can reduce demand without shifting purchases to another chip vendor.

The relevant market is therefore broader than Nvidia versus AMD. It includes custom silicon, model optimization, specialized inference chips, and changes in how customers allocate computing work.

Still, those routes should remain supporting context. The primary contest is Nvidia's actual earnings against Piper's projected earnings.

If AMD grows quickly while Nvidia still meets its forecast, Piper's thesis can remain intact. If Nvidia misses its earnings path without losing leadership, the valuation argument can still weaken.

This distinction matters because analyst targets often compress several assumptions into one output. Revenue growth, gross margin, operating expenses, taxes, and share count all influence projected earnings.

A strong revenue result does not guarantee the required earnings. Nvidia must also maintain pricing, manage product costs, and prevent operating expenses from absorbing too much growth.

Its 75.0% gross margin provides a substantial cushion. Yet the company already expects that figure to move toward 74.0% in the next quarter.

That change is small, but the direction deserves attention. Rack-scale products include more components and manufacturing complexity than individual accelerators.

Nvidia can generate more revenue per deployment while accepting a different margin structure. Investors must determine whether higher system value offsets the cost of delivering it.

Piper's Nvidia earnings forecast effectively assumes that the answer remains favorable. Competition becomes material when it changes that earnings conversion, even without changing the market leader.

What the Low Multiple Does Not Show

A low forward multiple can reflect genuine value, but it can also hide customer concentration, policy exposure, and uncertainty inside distant estimates.

Nvidia's latest filing shows that one direct customer represented 16% of second-quarter revenue. During the first half, three direct customers represented 16%, 15%, and 13% of revenue.

A direct customer can be a distributor, equipment manufacturer, cloud provider, model company, or systems integrator. The ultimate users behind those purchases can differ.

That structure makes concentration difficult to interpret from a headline figure. It still shows that a limited number of purchasing channels can materially affect quarterly results.

The company's latest filing also reports that five direct customers represented between 10% and 22% of accounts receivable.

Accounts receivable represents revenue recorded before customers have paid. Concentration in that balance can increase exposure to delayed deployments, financing conditions, or changing purchasing schedules.

Nvidia sometimes offers qualifying customers payment terms ranging from 90 days to one year. The company says these arrangements can support large data center builds.

Longer terms do not prove weak demand. They do show that financing conditions are becoming part of the infrastructure expansion.

The customer must still earn an acceptable return from AI capacity. Cloud providers need utilization, pricing, and related service revenue to support continued investment.

AI laboratories face a similar test. They need model subscriptions, enterprise contracts, advertising, or other income streams to finance rising compute requirements.

If end-user revenue grows slower than infrastructure spending, customers can adjust their orders. That adjustment might arrive after current backlogs have already supported several strong quarters.

Export controls create a second uncertainty. Nvidia excluded China Data Center compute revenue from its next-quarter forecast, reducing the immediate risk of a surprise within that guidance.

The longer-term effect is more complicated. Nvidia says export restrictions have effectively blocked it from China's data center compute market.

The company warns that controls can encourage customers to remove American semiconductors from product designs. They can also support larger ecosystems around foreign competitors.

Nvidia reported that shipments of Data Center Hopper products to China represented less than 1% of second-quarter Data Center revenue. That limits current exposure but does not measure the lost opportunity.

A forecast through fiscal 2030 must account for both outcomes. Nvidia might grow rapidly without China, or policy changes might reshape supply chains and competitive investment.

Product execution adds another risk. Nvidia's annual cadence requires timely transitions across chips, memory, packaging, networking, power systems, cooling, and software.

A bottleneck in one component can delay a complete rack. Customers buying integrated systems need every major part to arrive and operate together.

Supply constraints currently reinforce pricing and backlog. They can also prevent Nvidia from recognizing revenue when demand is strongest.

As production expands, the economics can change again. Greater supply can unlock revenue, but it can also reduce scarcity and expose the market's underlying price sensitivity.

The low fiscal 2028 multiple does not capture these risks separately. It represents their combined effect through a single earnings estimate and valuation ratio.

This is why the PEG ratio needs careful interpretation. Growth rates are estimates, while the share valuation reflects real capital committed today.

A PEG ratio near 0.3 looks compelling if earnings rise as expected. It becomes less meaningful if the growth estimate falls faster than the valuation multiple.

The metric also assumes that growth quality remains stable. Earnings supported by durable demand deserve a different interpretation from earnings supported by temporary shortages.

Piper's analysis is not necessarily too optimistic. It is simply sensitive to assumptions that investors cannot verify several years in advance.

Thirty Per Share Requires More Than AI Demand

Nvidia must translate industry demand into revenue, preserve strong margins, and control expenses for Piper's fiscal 2030 earnings estimate to work.

Revenue is the first requirement. A 47% compound growth rate means each year's result becomes the base for another large increase.

Nvidia already benefits from several simultaneous demand sources. Frontier laboratories train larger models, cloud providers rent accelerators, and enterprises add AI to existing software.

Inference expands the opportunity because deployed models process requests continuously. Agentic systems can increase usage further by performing multiple model calls during a single task.

However, demand for computing is not identical to demand for Nvidia revenue. Software improvements can lower the computing required for each unit of useful work.

Lower costs can stimulate more usage, creating a rebound effect. They can also allow customers to complete the same workload with fewer accelerators.

The outcome depends on elasticity, which measures how strongly consumption changes when costs fall. Piper's thesis assumes expanding usage will outweigh efficiency gains.

The second requirement is market capture. Nvidia needs enough product supply and sufficient system value to retain a dominant share of industry spending.

Its broad platform helps. Accelerators, CPUs, interconnects, switches, libraries, and deployment tools give customers one architecture for large AI clusters.

Yet complete systems expose Nvidia to more cost categories. Memory prices, advanced packaging, optics, networking, and power equipment can influence delivery schedules and margins.

The third requirement is pricing. Nvidia does not have to keep every product price unchanged, but revenue per unit of delivered computing must support the forecast.

New architectures generally offer more performance. Customers will compare that improvement with total ownership costs, including energy, facilities, and engineering time.

A system can become cheaper per unit of computing while generating more revenue for Nvidia. That happens when customers buy significantly larger deployments.

The fourth requirement is gross margin. Nvidia produced a 75.0% margin in the latest quarter, an exceptional result for a hardware-centered business.

Piper's earnings path becomes easier if margins remain near that level. It becomes harder if competition, product mix, or component costs reduce them.

A five-point gross-margin change would move billions in quarterly gross profit at Nvidia's current scale. That sensitivity grows alongside revenue.

Operating costs form the fifth requirement. Research and development spending is essential for maintaining the annual product cycle and supporting a growing software stack.

Nvidia recorded 7.1 billion in research and development expense during the quarter. Sales, general, and administrative costs added another 1.4 billion.

Those expenses grew slower than revenue, which created operating leverage. Operating leverage occurs when profit rises faster than sales because costs do not increase proportionally.

Piper's forecast implicitly requires that advantage to continue. Nvidia must invest enough to defend its platform without allowing costs to match revenue growth.

Taxes, investment income, and share repurchases also influence earnings per share. They matter, but operating performance remains the central driver.

This explains why a fiscal 2030 earnings figure should not be treated as a single prediction. It is the final result of several connected forecasts.

Revenue must compound near Piper's estimate. Market share must stay high, gross margins must remain strong, and expense growth must stay controlled.

A modest miss in one area might be absorbed by strength elsewhere. Simultaneous pressure on revenue growth and margins would be more difficult to offset.

The Nvidia earnings forecast therefore represents an execution corridor, not merely an addressable-market claim. Nvidia must stay inside that corridor across multiple product generations.

Three Signals Will Test Piper's Nvidia Forecast

The next evidence should come from reported revenue, margin behavior, and customer economics, in that order.

The first signal is Nvidia's fiscal third-quarter result against its 108.0 billion outlook. This is the closest test because management has already defined the expected range.

A result above that range would support the claim that supply remains the main constraint. It would also strengthen confidence in the company's fiscal 2028 growth outlook.

A result below the range would not invalidate the fiscal 2030 thesis by itself. It would force investors to identify whether timing, supply, demand, or customer spending caused the miss.

Data Center performance matters more than the consolidated headline. That segment contributed 89.0 billion in the latest quarter and remains the main earnings engine.

The second signal is gross margin as Vera Rubin production expands. Nvidia expects approximately 74.0% in the coming quarter, compared with 75.0% previously.

Investors should watch whether that movement reflects temporary ramp costs or a lasting change in system economics. A stable margin would strengthen Piper's earnings assumptions.

Continued compression would weaken the 2030 path unless faster revenue growth compensated for it. The explanation matters as much as the percentage.

Higher component costs differ from competitive discounting. Early production inefficiencies also carry different implications from a structural shift toward lower-margin products.

Management's disclosures about supply and product mix should help distinguish these factors. The most useful evidence will connect shipment growth with profitability.

The third signal is whether major customers continue increasing AI infrastructure spending while showing returns from that capacity. Orders alone do not establish durable economics.

Cloud providers need growing utilization and service revenue. AI companies need enough customer income or financing to support their computing commitments.

Continued customer spending alongside improving AI revenue would strengthen Piper's demand thesis. Spending cuts or extended payment periods would weaken it.

Customer concentration deserves attention within this signal. A broadening set of buyers would make the growth path more resilient.

Further concentration could still support strong results, but it would place more weight on a few investment cycles. Those cycles can change quickly.

AMD's progress also belongs within customer economics. Growing deployments would show that buyers want multiple suppliers, even if total AI demand continues expanding.

The key question is not whether Nvidia remains the largest vendor next quarter. It is whether the company keeps capturing enough value to support Piper's earnings curve.

That distinction prevents the analysis from becoming a simple market-share contest. Nvidia can remain dominant while producing returns below a distant forecast.

Piper's initiation has identified a real valuation argument. Nvidia's current growth, margin, product position, and backlog make that argument more than speculation.

Still, the apparent discount is conditional. The stock looks cheap on Piper's projected earnings, not on earnings that have already appeared in a filing.

Readers should treat each new result as a test of the forecast's components. Revenue growth, gross margins, and customer returns will reveal whether the model is holding together.

The Nvidia earnings forecast becomes more persuasive when those indicators improve together. It becomes less useful when a high headline growth rate hides weaker economics underneath.

For investors and industry watchers, the practical task is clear. Track the assumptions, compare them with reported evidence, and revise the valuation story when the evidence changes.