Asset Allocation Bi-Weekly – The AI Arms Race: Navigating the Divide Between Promise and Profit (10/6/25)

AI related investments have generated impressive results, but a look at how the big tech firms are channeling their current AI expenditures suggests it might be time for investors to put on the brakes at least a little. Confluence Associate Market Analyst and Certified Business Economist Thomas Wash joins Phil Adler to discuss why.

What is Confluence Podcasts?

Podcasts from Confluence Investment Management LLC, featuring the periodic Confluence of Ideas series, as well as two bi-weekly series: the Asset Allocation Bi-Weekly and the Bi-Weekly Geopolitical Report (new episodes posted on alternating Mondays).

Phil Adler: 00:04

Welcome to the Confluence Investment Management Bi-Weekly Asset Allocation report for 10/06/2025. I'm Phil Adler. AI related investments have generated impressive results, but a look at how the big tech firms are channeling their current AI expenditures suggests it might be time for investors to put on the brakes at least a little. Confluence Associate Market Analyst and certified business economist Thomas Wash joins us today to discuss why. Thomas, the title of your report this week is the AI arms race, navigating the divide between promise and profit.

Phil Adler: 00:43

Let's begin by talking about this divide. It appears that corporations in general across the board aren't getting significant returns yet on their AI investments. What's the problem?

Thomas Wash: 00:55

I think what you're asking is a core issue, and it highlights a critical distinction. There's a major difference between selling AI and using AI effectively. Tech firms selling AI tools are seeing massive returns as demand is sky high. However, for the corporations implementing these tools, the results have been, you know, somewhat disappointing. A recent MIT study confirms that most are not seeing significant boost in productivity or revenue.

Thomas Wash: 01:22

The problem isn't the technology's potential, it's a lack of strategic integration. Companies are rushing to adopt AI without the necessary expertise, process, or clear goals. This leads to a phenomenon known as AI debt, where the hidden costs of implementation, integration, and maintenance end up being far greater than the initial investment eroding any potential return.

Phil Adler: 01:44

Well, has this lack of significant returns changed the way many companies are now investing in AI?

Thomas Wash: 01:52

Yes. And I think that's a key question. And the lack of immediate massive returns is already shifting the AI investment landscape. Our overall corporate adoption remains high. The pace of investment amongst larger firms has started to slow.

Thomas Wash: 02:06

This deceleration isn't just poor execution. It signals a change from broad experimental deployment to a more deliberate strategy. It looks like companies are now stepping back to focus more on governance, coordination, and building the necessary internal infrastructure to ensure AI models can deliver measurable, scalable value across the corporate structure.

Phil Adler: 02:27

Let's turn now to how the major technology companies, have driven this current bull market, are spending their piles of cash. What does the evidence show?

Thomas Wash: 02:38

Well, the evidence shows that the major technology companies are allocating vast reserves of cash toward capital expenditures, driven by the demands of generative AI. This represents a fundamental shift from their historical capital light business model to a capital heavy one. The funds are being primarily channeled into AI infrastructure, specifically the acquisition of specialized hardware like GPUs and the build out of large scale data centers necessary to power complex AI algorithms.

Phil Adler: 03:05

I wonder about the depreciation of these companies' cash reserves. Is this something investors should be concerned about?

Thomas Wash: 03:13

Whether this is a concern depends on the investor's time horizon. Cash reserves are being depleted primarily to fund massive capital expenditures on AI infrastructure. This represents a clear trade off. Investors must accept lower immediate shareholder returns, meaning fewer dividends and buybacks in exchange for what company hopes will be a substantial long term competitive advantage and superior revenue growth enabled by the new infrastructure. If the expected AI return fail to materialize, the reduced cash reserves will become a significant problem.

Phil Adler: 03:44

Thomas, let's back up a bit for a moment. Since it's obvious the big companies are investing in AI infrastructure, maybe we should ask the question, what exactly constitutes AI infrastructure?

Thomas Wash: 03:56

You know, that's a that's a great question. Think of AI infrastructure as the modern factory for intelligence. It's not one single thing, but a stack of essential components. At the foundation, you have the compute level, the raw power, which is primarily advanced GPUs housed in massive data centers. Then you have the software and platform layer, the tools and services like those from the major cloud providers that allow developers to build, train, and manage those complex AI models.

Thomas Wash: 04:23

Finally, and crucially, it includes the data management systems that fuel the entire operation. So it's the entire ecosystem from the physical chips and building to the platforms that make them usable.

Phil Adler: 04:36

Are these AI infrastructure companies reaching the point or coming closer to the point where they're overly dependent on on a relatively few sources of revenue?

Thomas Wash: 04:47

You know, that that is a very perceptive point and is becoming a central concern. While tech companies are posting impressive revenues, the growth appears heavily concentrated among a handful of major players. This creates a dual risk. First, it suggests a potential ceiling on the market if the technology remains prohibitively expensive for the broader ecosystem of small and mid sized businesses. Second and more acutely, it creates a systemic vulnerability.

Thomas Wash: 05:12

This entire revenue model for these tech firms could be exposed to a sudden pullback in spending from just acute a few key clients, making them susceptible to an industry specific shock.

Phil Adler: 05:21

You also make the point in your report, Thomas, that it looks like a major portion of AI funding is circulating among relatively few companies. Is this another reason for some caution?

Thomas Wash: 05:34

Absolutely. It's a primary reason for caution. We're seeing a self contained cycle where a great deal of AI investment and revenue is circulating among a small ecosystem of tech giants. They're basically investing in each other's infrastructures and capabilities. Now the risk is that this creates a bubble of expectation.

Thomas Wash: 05:54

The market is valuing these companies as if their technology is already being widely adopted across the global economy. But if the adoption beyond their inner circle remains slow and the pool of viable clients doesn't expand significantly, these companies will struggle to justify their current valuations when it's time to deliver concrete earnings.

Phil Adler: 06:12

Let's turn to the role of the federal government. You suggest in your report that the government is boosting the AI investment story. Has tax legislation encouraged investment?

Thomas Wash: 06:23

Yes. Recent tax legislation had played a significant role in encouraging investment, particularly in the tech sector. Key provisions such as enhanced r and d tax credits, accelerated depreciation for capital expenditures, and other investment friendly policies have made it more attractive for companies to continue spending on innovation and infrastructure. These incentives are likely a major factor behind the sustained capital investment we're seeing, especially in AI and related technologies. As a result, we believe these legislative supports will help extend the current investment boom for at least the next couple of years.

Phil Adler: 06:55

Looking at the Intel investment, which was in the news recently, does the fact that the the federal government is willing to take a stake in the company suggest that more such investments are on the way and and we can be encouraged that momentum will continue and our AI investments are safer as a result?

Thomas Wash: 07:12

Yes. The government's recent involvement with Intel as well as its collaboration with the Stargate Fund is a strong signal of its commitment to supporting The US tech industry. These moves reflect a broader strategy to strengthen the domestic innovation, particularly in critical sectors like semiconductors and AI, by backing companies that are reshoring operations and investing in US based infrastructure. This kind of public private partnership not only provides financial support, but also creates a politically favorable environment for further expansion. As a result, it suggests that similar investments could follow reinforcing momentum in the sector.

Phil Adler: 07:47

Well, let's get to some bottom line questions for investors. Do you think, first of all, that there is enough uncertainty on the AI investment front to derail the bull market?

Thomas Wash: 07:59

So while we acknowledge that the tech sector faces some vulnerabilities, we remain confident that the current bill market has at least some more room to run. Our optimism is based on two key factors. First, many leading tech companies maintain relatively healthy balance sheets, providing them with the financial flexibility to continue investing and innovating. Second, historical trends show that bull markets typically last around five years on average. Given that the current rally began around October 2022, more or less, it's reasonable to believe that the momentum could persist for some time.

Phil Adler: 08:35

So what's your advice for stock investors?

Thomas Wash: 08:39

Broadly speaking, you know, given the market's heavy concentration in large cap tech stocks, we believe investors should be mindful of the elevated risk of a potential correction in high growth names. To manage this risk, we recommend maintaining balanced exposure by including value stocks in your portfolio. These value oriented stocks tend to exhibit lower volatility and have historically offered greater resilience during periods of market stress or when growth stocks come under pressure. This diversification can provide a defensive cushion and help smooth returns across market cycles.

Phil Adler: 09:12

The final question. Within this value group, does Confluence currently favor energy and power companies, which AI firms are counting on to help meet their needs?

Thomas Wash: 09:22

Yes. We maintain a selective and cautious outlook on the broader energy sector. While the traditional energy market, particularly oil and gas, has faced headwinds, we see a compelling opportunity in utilities and power providers positioned to supply the massive electricity demands of data centers. However, the growth is not without risk. The surge in energy demand could lead to higher household utility bills and regional grid constraints, increasing the potential for political and regulatory pushback.

Thomas Wash: 09:51

Therefore, our focus is on companies with the capacity to scale sustainably and navigate this complex landscape.

Phil Adler: 09:59

Thank you, Thomas. The title of this week's report is the AI arms race navigating the divide between promise and profit, and you can find a link to the written report on the Confluence webpage, confluenceinvestment.com. Our discussion today is based upon sources and data believed to be accurate and reliable. Opinions and forward looking statements expressed are subject to change without notice. This information does not constitute a solicitation or an offer to buy or sell any security.

Phil Adler: 10:26

Our audio engineer is Dane Stoll. I'm Phil Adler.