Nvidia's Growth Is Supported By Hyperscalers' Capital Investments
Feb. 18, 2025 6:52 AM ETNVIDIA Corporation (NVDA) Stock, NVDA:CA StockNVDA, NVDA:CA19 Comments
Michael Del Monte
4.36K Followers
Play
(9min)
Summary
Nvidia's q4'25 earnings have a strong potential to outperform driven by strong data center growth and hyperscalers' continued investment in compute capacity.
Hyperscalers' total capital investments may grow by nearly 40% in CY25 to $313b; investments will primarily go towards compute and data centers.
The release of DeepSeek-V3 models emphasizes the need for AI model optimization; I do not believe this will directly translate to lower demand for GPUs.
Enterprises are expected to bring GenAI into production in CY25, driving the need for data center computing capacity.
Powering Intelligence: AI Over a Grid of Technology
J Studios
Nvidia (NASDAQ:NVDA) is set to report q4’25 earnings on February 26, 2025, after market close. With the recent news of the release of the DeepSeek-V3 models emerging, the market has been questioning the viability of the growth trajectory for compute capacity, potentially leading to a paradigm shift in investments for Nvidia’s GPUs. Despite the narrative, I believe two important factors emerged from this. The first is that we are in the phase of model optimization, leading to more efficient compute. The second is that compute capacity still needs to scale to meet enterprise demand, as denoted by the hyperscalers’ investment outlays. Given the optimistic spend outlay, I am reiterating my BUY rating for NVDA shares with a price target of $234/share at 28x eFY26 price/sales.
The Case For Nvidia
The recent release of DeepSeek-V3 GenAI models took the market by storm, leading the investment community to question the viability of the high level of compute costs associated with running existing models. Though DeepSeek may not be the gold standard of AI models in terms of performance, the application advanced down the cost curve beyond any competing model, pushing the cost of compute to a new low.
One of the factors that might need to be considered is scalability. The theme across the hyperscalers’ earnings performance has slightly changed from: “Demand continues to outstrip supply.” The new theme is that growth is constrained by their ability to build data centers and deliver compute capacity. Though these two statements are similar, they are somewhat different.
Each of the hyperscalers addresses the concern that DeepSeek (DEEPSEEK) brought to the market and welcomed the model with open arms. In fact, Microsoft and Amazon are all hosting DeepSeek R1 on their respective platforms. Though it may have been a positive spin on disruptive news, the general theme across the hyperscalers was that DeepSeek isn’t necessarily going to disrupt the need for additional compute capacity. The primary conclusion was that DeepSeek’s models are seen as the next phase in AI inferencing, in which models will need to be optimized to function more efficiently. Though the release of DeepSeek’s models may bring forward some concerns relating to scaling compute capacity, I do not believe the existing data center footprint is developed to the point of turning to economies of scope. If anything, I expect that the emergence of these models will ignite the necessity of domestic AI developers to accelerate growth and development to maintain their leadership in AI.
As for building a case for investing in Nvidia, not a single hyperscaler reduced its capital outlay for compute and data centers. Borrowing a chart from my recent report covering Alphabet, growth in cloud computing remains strong across the hyperscalers.
Corporate Reports
Corporate Reports
What’s most impressive is the capital outlay. On a trailing twelve-month basis, the 3 CSPs invested $186b in expanding their compute capacity.
Looking ahead to eCY25, I’m forecasting the capital outlay to grow by nearly 40% on a year-over-year basis, up from $225b. Beyond eCY25, I’m anticipating that growth will normalize and level out at this heightened rate.
Corporate Reports
Corporate Reports
It should be noted that these figures are my expectations alone. Beyond eCY25, the estimates are based on the 24% CAGR estimate for data centers through 2028. Meta Platforms (META) has suggested a range of $60-65b for eFY25, Microsoft (MSFT) guided that q2’25 set the tone for the e2h25 capital outlay, Amazon (AMZN) set the pace of at least $100b for eFY25, and Alphabet (GOOG) (GOOGL) is expecting to deploy $75b in eFY25.
Unless demand for compute capacity softens, the hyperscalers have no plans to slow down investing in new capacity. Management has made clear that capital deployed will remain dependent on the demand environment.
Driving this further, Oracle Corp. (ORCL) is partnering with OpenAI in Project Stargate, a $500b project for advancing AI in the US. Oracle alone has pinned 100 data centers for future development, with a site as large as 1.60GW in capacity.
In the physical AI space, Tesla recently completed its 50,000 H100 cluster at its Gigafactory Texas facility in Austin that will be utilized for autonomous driving. Elon Musk suggested that the compute capacity for developing its Optimus humanoid robot will need to be 10x larger in order to properly develop a fully autonomous, general-purpose robot.
One major factor that must be considered is the infancy of AI at the workplace. Gartner is forecasting GenAI to move beyond the proof-of-concept phase and into production in 2025. Gartner is also forecasting server sales to triple from 2023 to 2028. Though much of the growth may be tied to the CPU server refresh cycle, I anticipate that a good proportion of the growth will be tied to GPUs.
Nvidia Financial Position
Corporate Reports
Corporate Reports
I’m forecasting top-line growth to remain strong at 73% on a year-over-year basis, bringing total revenue to $38b with an adjusted EPS of $0.85/share. This will be driven by an 84% year-over-year growth rate for its Data Center segment following the ramp-up realized in FY24. Though I’m forecasting growth to begin tapering in eFY26 relative to rates experienced in 2h24-1h25, I believe growth will remain durable given the high levels of investment anticipated by the hyperscalers and enterprises.
Advanced Micro Devices (AMD) reported exceptional strength in its Client segment, growing by 58% in q4'24 on a year-over-year basis. This may potentially act as a precursor for Nvidia going into eq4’25.
Risks Related To Nvidia
Bull Case
Capital investments by the hyperscalers geared towards growing compute capacity remain durable and growing for eCY25. Enterprise use of GenAI remains a novel experience, potentially leading to a larger upswing in demand for compute capacity to support AI inferencing. Nvidia may also realize growth in its Gaming segment as more workloads are executed at the edge.
Bear Case
Growth may begin to taper sooner than expected if demand for Nvidia’s GPUs doesn’t remain as robust as anticipated. If the hyperscalers begin pulling back on their capital outlay, Nvidia’s growth may slow.
Valuation & Shareholder Value
Corporate Reports
Corporate Reports
NVDA shares continue to trade at a significant premium to its peer semiconductor designers.
Seeking Alpha
Seeking Alpha
I believe NVDA shares’ trading premium is justified given the company’s high-growth trajectory as well as its industry-leading profitability.
Seeking Alpha
Seeking Alpha
Seeking Alpha
Seeking Alpha
Using an internal valuation model based on my revenue forecast and NVDA shares’ historical trading premiums, I am reiterating my BUY rating for NVDA shares with a price target of $234/share at 28x eFY26 price/sales.
Corporate Reports
Corporate Reports
Using the model: the valuation table above references my financial forecast in the firm’s “financial position” section and ties it to the stock’s historical trading premiums. The trading premium array is derived through the normal operating cycle, with the blue-sky scenario being the stock’s peak multiple and the gray-sky scenario being the lowest point. The target multiple aims for the midpoint, or the most likely trading range for the company’s stock. The trading multiples from there are set to a probability factor based on the likelihood of the stock trading at that premium based on its historical presence. From there, the trading multiple is tied to the probability factor to derive its relative market cap and relative multiple.
This article was written by
Michael Del Monte
Market Data 
Markets 
AI
