The Visible Alpha AI Monitor August update: What’s next for AI?

The Visible Alpha AI Monitor aggregates publicly traded US technology companies, providing a comprehensive measure of the current state and projected growth of the core AI industry. This encompasses the AI-exposed revenue for companies that are building AI infrastructure and capabilities for both enterprises and consumers.

Investors can use the Visible Alpha AI Monitor to generate new ideas to capture growth emanating from the core AI industry, and to evaluate the potential AI exposure of technology stocks in their existing portfolios. We have identified specific line items that capture potential growth of AI-related revenue that are available in the Visible Alpha Insights platform.

Key Questions for H2 2025:

1. How will President Trump's policies impact the AI trade?
2. Nvidia dominated both the AI Monitor’s stock performance and upward revisions to AI-exposed revenue in 2024. Will this continue throughout 2025?
3. Will AI agents and AI laptops help to drive broader adoption in the second half of 2025?

Introduction: Agents?

The generative AI (GAI) trend gained further momentum in 2025 as Cloud Service Providers invested heavily in capex to transition datacenters to accelerated computing. Visible Alpha observed that companies attempted to make a greater push with GAI into their organizations to improve efficiency and enhance the client experience. While the usage and output quality of ChatGPT and its peers improved in 2025, there have been no game-changing new applications leveraging GAI.

Over the past few years, innovation in the chip and model has benefited NVIDIA Corp. (NASDAQ: NVDA). However, there has been less innovation at the application level to drive broader adoption of GAI with end users, leading to a more competitive business model. The verdict is still out on the magnitude of the productivity impact GAI may yield for businesses and individuals. Could these productivity benefits help to deliver stronger fundamentals and earnings growth, or will they be merely anecdotal?

So far this year, there have been signs of GAI adoption happening more broadly in enterprises with the introduction of AI agents into role-specific workflows. AI agents seem to enable domain and persona-specific workflows to complement human roles in an organization. For example, a firm may have a unique AI agent for Research, Security, Analytics, Sales, and Customer Service to complement the human work done in these functions. We want to see how firms leverage these agents and the potential direct or indirect impact on revenue and cost in the longer term. The challenge is getting the persona and the domain-specific output correct.

Based on comments in SIGGRAPH 2025 in August, agents will be a gateway to Physical AI by controlling digital tools and providing the groundwork for controlling physical tools. According to Aaron Lefohn, VP of Research at Nvidia’s Real Time Graphics Lab, AI is transforming every stage of the graphics pipeline, while graphics is simultaneously teaching AI about 3D worlds. By merging neural rendering, generative AI, and advanced simulation, Nvidia aims to make city-scale, fine realism achievable in real time - paving the way for both richer virtual worlds and more accurate physical AI training environments. According to Sanja Fidler, VP of Research, Special Intelligence Lab at Nvidia, Physical AI represents a massive upcoming market, and simulation is the key to unlocking it. Real path tracing and neural shading are key to simulation to support the testing and training of models.

GAI enhances this process by augmenting environments with rare or hazardous events (e.g., a burning car), changing lighting or weather, and inserting new objects with correct shadows and reflections. This makes it possible to create endless environment variations from a single real-world capture, exposing robots to edge cases they might otherwise never encounter safely. This domain-specific, detailed, complex testing is critical for client-facing and internal GAI output to be precise and impactful within enterprises. (see the goals, objectives, and methodology of the AI Monitor at the bottom of this page).

Visible Alpha AI Monitor

January to August 2025 performance summary

The Visible Alpha AI Monitor universe of 61 publicly traded US companies is 79% weighted to the 10 largest companies, with the remaining 21% dispersed among 51 companies. On a market-cap-weighted and AI-exposed revenue-weighted basis, the Visible Alpha AI Monitor was driven by stock price outperformance (vs. the S&P 500 index) of the largest companies this year. Smaller company performance (vs. the S&P 500 index), especially on an equal-weighted basis, has lagged in 2025, but the performance has started to broaden out on a stock-specific basis. On an equal-weighted basis, the AI Monitor generated an overall lower return when compared to the market-cap and AI-exposed revenue-weighted aggregations this year, driven by the drag of a lower return generated by the smallest names. In 2025 so far, the larger market cap names continue to drive outperformance.

Top 10: Nvidia continues to dominate expectations

Based on an analysis of the 10 largest players, 2026 revenue forecasts for AI-exposed revenue segments increased by a total of $27 billion since January 2024. Nvidia accounts for nearly $14 billion of this increase, Super Micro Computer for $3 billion, and Oracle for a further $3.5 billion. Slightly lower expectations for AWS and Dell drove the small downward revisions.

Smaller contributors

The remaining 51 companies could serve as a good place for investors to discover ideas by surfacing expanding new players. Although smaller companies in aggregate have not performed as well as the Top 10, there have been some clear outperformers relative to the composite. Revenue growth expectations are very mixed, with strong double-digit revenue growth expected at some firms, and estimate declines anticipated at others. These dynamics may help investors identify emerging trends in the space.

Between January and August 2025, we have seen that to be true with Palantir Technologies Inc. (NASDAQ: PLTR), Seagate Technology Holdings PLC (NASDAQ: STX), and BigBear.ai Holdings Inc. (NYSE: BBAI). Each delivered a strong outperformance (vs. the Russell 2000), which helped position them longer term as the possible up-and-comers in the space. In sync with these outperformers, 44% of the smaller companies in the AI Monitor outperformed the Russell 2000 return of 1% so far in 2025. This strength may imply that the fundamentals and valuations of the smaller firms may be starting to regain momentum in the tariff era.

What is moving the Al Monitor?

From January to August 2025, seven out of the 10 largest Al-exposed revenue generators drove strong outperformance, while only 44% of the smaller-cap Al stocks outperformed the small-cap index. Al-exposed revenue should grow by close to $520 billion, from $472 billion at the end of 2023 to $991 billion at the end of 2026, driven overwhelmingly by the Top 10 largest companies in the Al Monitor, and 38% coming from Nvidia alone.

Nvidia: Al GPU TAM?

From the end of 2023 to the end of 2026, consensus expectations for Nvidia's AI-exposed revenue were revised up by an aggregate $535 billion. These upward revisions contributed significantly to both the Al-exposed revenue concentration and the stock performance of the Al Monitor.

In 2024, there was a further $200 billion in upward revisions in addition to the initial $200 billion made in 2023 to Nvidia's Data Center revenue estimates. This optimism was driven by continued heavy capex investment by the cloud service providers to support transforming datacenters to accelerated computing to support Al applications. Capex spending from Meta Platforms Inc. (NASDAQ: META), Alphabet Inc. (NASDAQ: GOOGL), Amazon.com Inc. (NASDAQ: AMZN), Apple Inc. (NASDAQ: AAPL) and Microsoft Corp. (NASDAQ: MSFT) should exceed $350 billion in 2025, up significantly from $159 billion in 2023. In 2025, there have been further upward revisions for Nvidia, but their pace and magnitude have been smaller. Coming into the Q2 earnings release, the pace of revisions to Data Center profitability has moderated for the first time in nine quarters, due to the ramp of Blackwell. Questions remain about how much of the capex spending surge has already been captured by the market and if estimates may see further upward revisions. Consensus revisions seem to have plateaued, as Blackwell ramps, but the AI GPU total addressable market appears to be expanding.

Expectations continue to move higher, but at a smaller pace

Dell Technologies Inc. (NYSE: DELL) revenue estimates for the Al-exposed Infrastructure Solutions Group rose nearly $30 billion in 2024 but are flat in 2025 to-date. Dell is due to report earnings at the end of August and provide guidance about the second half of 2025. Will Al servers and the potential for a replacement cycle fueled by must-have next-generation Al laptops show an uptick in H2 2025? 2025 and 2026 Google Cloud were revised up in 2024 by $26 billion, but only $2.8 billion in 2025 to-date, implying a more in-line outlook. Intel started to emerge from a period of weakness, with estimates increasing by $1.3 billion for 2026.

Microsoft's Azure business has not moved much, although Microsoft did recast the accounting of its segments, which impacted the Azure revenue line. In addition, the company created a new line called Al Services to showcase this emerging revenue line. In FY 2024, this business generated $4 billion and it should ramp up growth and generate over $37 billion by the end of FY 2027, based on Visible Alpha consensus.

Last year, Super Micro Computer Inc. (NASDAQ: SMCI) experienced significant volatility due to issues with its auditor. For 2025 and 2026, the Al-exposed revenue estimates have increased by over $30 billion, as analysts gain more visibility into their growth. Given how rocky 2024 was for the company, it is interesting to see the rebound in the stock in 2025.

What about Apple?

In addition to the Top 10, we are monitoring the potential Al revenue trends at Apple. The company released Apple Intelligence last summer and has embedded many new Al capabilities in its latest iPhone models. These product updates have not garnered much excitement with users, and there are concerns that the new Al functionality has not been enough to make users want to upgrade their older phones. In the latest earnings release, the company committed to increasing investments for Al. There are questions about the strategy and whether Apple may opt for a large acquisition in the space. With over $120 billion of cash on the balance sheet, it could make a big move.

Expectations for 2025 iPhone units have come down to 232 million from 237 million last year, in part due to lower growth in China. It will be interesting to see and hear what new updates Apple may reveal in H2 2025 that could get the market excited and fuel upgrades. For now, the next iPhone upgrade cycle remains slow.

Regulatory backdrop: Impact from President Trump's new policies

President Trump wasted no time and is focusing on securing Al leadership for the US. Biden's rhetoric around regulation, governance, and oversight of Al has been replaced with Trump's move to establish new Al infrastructure initiatives in the US, in tandem with new tariff policies.

In early 2025, Trump launched the Stargate Project, an Al infrastructure company that plans to invest $500 billion over the next four years to build out new Al infrastructure in the US. Stargate will be initially financed by SoftBank, OpenAl (Microsoft), Oracle Corp. (NASDAQ: ORCL) and MGX. Speaking with President Trump, Oracle CTO, Larry Ellison, Softbank CEO, Masayoshi Son, and OpenAI CEO, Sam Altman, outlined the ambitious goals of Stargate and their initial commitment of $100 billion and the subsequent $400 billion of financing over the next four years.

In July 2025, the approach was replaced on the ai.gov site with President Trump's Al Action Plan. There is a clear change in direction on a few key initiatives. The new administration seems to be more focused on building and securing the infrastructure, instead of trying to regulate Al. Stanford University released an update in April 2025 to its Al Index. The trajectory of the proposed regulations suggests we are unlikely to see more regulations in 2025. Most of the global Al pending is happening in US Defense.

Final thoughts

Given Trump's support of AI and the expansion of AI infrastructure in the US, cloud service providers (CSPs) will remain front and center. In 2025, Oracle and Palantir led, driven by strong demand for Al infrastructure and security solutions. Will this continue for the rest of 2025?

Debates are emerging about how best to create scalable enterprise applications for GAI. For many firms, it is unclear how they will bring GAI to enterprises and grow the impact of Al exposure in their business models. The importance of specializing and bringing domain expertise to smaller models will be increasingly critical to the success of these tools. In 2025, we are watching the pace at which companies are enabling innovation with Al in enterprises. Some skepticism is emerging about the actual use cases, real impact, and ROI. In H2 2025, Al agents and Al laptops may be compelling mechanisms for fostering broader adoption and driving specialized applications in enterprises, but the verdict on these new Al tools is still out.

We will watch the evolution of technology partnerships with the US government. Initiatives, like the Stargate project partnership between ARM, Nvidia, Oracle, Softbank, and OpenAl (Microsoft) could drive growth in sales and earnings for the key players. There may be a few new companies in the Al space that emerge as beneficiaries of the significant Al infrastructure growth in the US. While the performance of Al-related stocks has been choppy in 2025, Al demand remains strong for the key players. This seems to be partially driven by the US government's continued focus on AI for defense and national security.

AI Monitor goals and objectives

The objective of the Visible Alpha AI Monitor is to show the investment community which companies are likely to drive AI. As the world embraces AI and its applications to enterprise workflows and our daily lives, there are big questions about how AI’s impact on company business models will unfold over the next 3-5 years. AI can potentially free people from tedious grunt work to enable more focus on critical workflows that require human creativity and analysis.

A primary goal of the Visible Alpha AI Monitor is to show which US companies and specific line items we are watching as the embryonic AI theme emerges across company fundamentals and begins to scale broadly across the economy. We are monitoring how AI may be reflected in the numbers, and which companies may be benefiting more or less. This universe attempts to be comprehensive and to show investors the dynamics of both the large and smaller US players. Additionally, it aims to help investors identify new names that may be smaller and less covered but potentially growing and emerging more quickly.

AI Monitor methodology

Using Visible Alpha’s comprehensive database of detailed estimates pulled directly from sell-side analysts’ spreadsheet models, we have assembled an aggregation with a universe of 61 publicly traded companies that are contributing to the infrastructure and broad scaling of AI capabilities. This monitor aims to provide a current and future snapshot of where AI-related revenue is and is not growing across each of these 69 companies, particularly the 10 largest.

We have aggregated the revenue of specific business segments at firms that are driving the wider AI trend. For larger firms, we have attempted to pinpoint where in their revenue model AI is driving growth. For some smaller firms, we are simply incorporating 100% of revenue. The AI-exposed revenue lines we identify are intended to be used as a proxy for monitoring the growth of each company’s AI business. Given both the lack of discrete company disclosures and how intertwined AI and conventional technologies and services can be, these lines should not be taken as exact quantifications of AI revenues, but are, we believe, the best systematic approximation available.

The AI Monitor provides three measures of stock performance for its universe. These metrics are meant to show the returns of various weighting schemes. The returns are calculated on both an equal-weighted and market-cap-weighted basis. The universe performance of the AI Monitor is also weighted based on AI-exposed revenues and calculated in aggregate. From 2024, the return calculations were standardized, and market-cap-weighted now reflects year-over-year changes.

For Visible Alpha subscribers, details of these companies can all be found within the Visible Alpha Insights platform. Each company included in the monitor has coverage by at least four sell-side analysts. In addition, given the quickly evolving state of the AI space, these line items are subject to change and may shift significantly over time. We plan to refresh the data on an ongoing basis and provide regular updates.

This article was published by Visible Alpha, part of S&P Global Market Intelligence and not by S&P Global Ratings, which is a separately managed division of S&P Global.