Stock Talk - Technology Adoption Cycles: The AI Capex Spending Tsunami

Episode Date: November 26, 2025

Over the past decade, the “Magnificent 7” stocks grew rapidly by operating relatively asset-light businesses. But today, these same companies are allocating unprecedented amounts of capital to chips, servers, data centers, electricity, real estate, and AI infrastructure. My goal in this video is to give you clear context around these spending trends, how they compare to past investment cycles, and why investors are raising new questions about sustainability and long-term returns.   What if I told you the shape of today’s economy might explain why some Americans feel prosperous while others feel like they are falling behind? In this episode of Stock Talk, I break down why economists are calling this a K shaped recovery and what that really means for consumers, workers, and long term investors.   Over the years, commentators have tried simplifying economic cycles with shapes like V or W. But what we are experiencing today looks very different, and the divergence in outcomes between groups is becoming clearer every month.   In this week’s discussion, I share what I’m hearing from hundreds of corporate earnings calls, how spending patterns are shifting, and what these changes might signal for markets going forward. My goal is to provide a calm, educational perspective so you can better understand the environment you are planning and investing in.   About Chris Perras, CFA®, CLU®, ChFC®, Chief Investment Officer: As CIO, Chris is the lead investment strategist and director of research at Oak Harvest Financial Group. Chris develops the firm's core market outlook, putting his decades of experience and expertise to work for our clients. He hosts Oak Harvest's podcast, "Stock Talk," available on the website with new episodes each week. He completed his undergraduate studies at Georgia Tech, and went on to obtain an MBA from the Harvard Business School. Driven by a desire to maximize his knowledge and skill set, he acquired financial planning and investment management qualifications, becoming a Chartered Life Underwriter (CLU®), a Chartered Financial Consultant (ChFC®), and a Chartered Financial Analyst (CFA®).   Stock Talk is a weekly vlog/podcast dedicated to discussing the Oak Harvest Financial Group Investment Team's perspective on what's happening in the market. Hosted by Chief Investment Officer Chris Perras, each episode brings you our views on stocks, the market, and the economy — with a little education thrown in for good measure. Listen each week and help stay connected to your money!   Do you need a retirement plan that goes beyond allocating funds to truly fit your needs? We can help you create a retirement life plan customized for your retirement vision and legacy. Call us at 877-896-0040 or fill out this form for a free visit: https://click2retire.com/lets-connect   Important disclosures: Content of Oak Harvest podcasts expresses the views of the speaker and is for informational purposes only. Oak Harvest believes that any data, articles, or information cited are reliable at the time of creation, but does not warrant any information contained herein to be correct, complete, accurate, or timely. References to third-party analysts should not be seen as an endorsement of their views or recommendations, and you should do your own research before investing. The views and opinions expressed herein may change without notice. Strategies and ideas discussed may not be right for you, and nothing in this podcast constitutes personalized investment, tax or legal advice, or an offer or solicitation to buy or sell securities. Indexes such as the S&P 500 are not available for direct investment and your investment results may differ when compared to an index. Any specific portfolio actions or strategies discussed will not apply to all client portfolios. Investing involves the risk of loss, and past performance is not indicative of future results.

Okay, investors, before getting into this week's topic, which is making front page financial headlines week after week, that's the AI technology capital expenditure tsunami, the team here hopes you had a peaceful and thankful Thanksgiving yesterday. So over the past 10 years, the stocks of the seven large technology companies, Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla have performed so well and grown by such big size, they've been grouped into a universe known as the Magnificent Seven. or Mag 7 for short. So the Mag 7 stocks as a group have significantly outpermed the rest of the S&P 500. Investors since 2015, these seven stocks have combined to return almost 27.5% per year, largely due to the extraordinary high revenue growth, high margins,

and high operating profit per employee on the back of being relatively asset light. However, recently, the Mag 7 cohort has ramped up capital spending significantly, And this spending habit shows no sign of slowing as they're projected to continue to spend aggressively on semiconductor chips, racks, servers, and electricity and real estate into the near future. Investors, as they spend more, they shift from asset light to more asset heavy and the risk of over investment and lower marginal return on invested capital grows with it. Meta, Microsoft, and Alphabet are each set to spend between 21 and 35% of their revenue on CapEx and the next few years. Combined, the seven, our estimated spent almost $400 billion per year the next few years. If Jetson Wong from NVIDIA is correct, that number might be closer to a half a trillion per year. Big Tech's AI spending is currently so large that it's helping the broader economy,

accounting for an estimated half of U.S. GDP growth so far this year. For the Carson Group, a chart on incremental GDP from tech spending. As investors, you can see this is spend level is now currently above. incremental GDP contribution during the late 1990s.com capex boom so why the recent concerns well past infrastructure spending booms have taught investors that the path of sustainable revenue is often set by periods of poor investment and poor returns so-called booms and busts whether it was the railroad build-out in the early 1900s car manufacturing boom after world war two the initial boom in personal computing penetration in the late 70s and early 80s or most

recently, the dot-com fiber capacity buildout in the late 1990s, or even more recently, electric vehicle boom post-2010. According to Mary Meeker, who sat at the center of the dot-com boom, generative AI penetration is reaching prior penetration rates at an unprecedented pace. Chat, GPT, hit 800 million weekly active users in just 17 months. Meeker said it took the internet nearly 25 years to get to that many users. PCs took 20 years to reach. reach half of U.S. households. Desktop internet took 12 years, and mobile internet took six years. And AI appears to be reaching 50% penetration here in the United States in just three years. Bestors, here's a great craft from Forbes for prior technology advancements and their adoption

curves over the last century. These buildouts and technology advances provide examples of capital investment cycles. Initially, the excitement around new technology spurs firms to make massive capital investments. Investors reward these accelerating outlays with soaring stock prices, encouraging even more investment. In the late 1990s fiber buildout, telecoms and startups laid 80 million miles of fiber optic cable. Unfortunately, over the next 5 to 10 years, 85% of that cable laid unused, and the cost of bandwidth transmission on these lines fell by minus 90%. User demand didn't keep up with a new supply coming online, and that caused prices to crater, saddling the industry with years of excess capacity.

Corporate values collapsed in those that took on debt to finance the buildout. Many had to file for bankruptcy. However, that minus 90% reduction in fiber costs fueled successes of others like Google's YouTube, Netflix, Facebook, whose business models flourished on the back of plummeting internet transmission costs. Investors up until about two months ago, 2025 investors looked at these accelerated AI KAPX investments favorably. However, recently, the tone has changed amongst a flurry of debt-financed expanses, off-balance-sheet commitments, and unfinanced gargatuan CAP-X future spending plans by Open AI and others. While the topic of accelerating AI CapEx spending from a sustainable return on investment capital and accounting treatment standpoint has been discussed throughout their technology investment circles for two to three years now,

the last four to six months and the voice, the social platform, an investor Michael Burry, have recently brought the accounting discussion to the forefront of many investors' minds. Here's the progression of the depreciation schedules for a few of the Mag 7 from public records as compiled by Mr. Burry. This topic currently sits in the crosshairs of short sellers and those currently negative on AI spending. As one can see, in general, all of these companies have lengthened their depreciation schedules the last five years, some nominally like Oracle, others rather substantially like Meta, Google, and Microsoft. Investors, while this does little to change a company's free cash flow, it does have the effect of increasing their reported earnings as depreciation is a non-cash charge on the income statement.

Appreciation charges try to estimate the useful life in value, longer-lived, fixed capital investments. Burry's argument is that these companies are overstating earnings significantly because, as opposed to longer-useful lives, lower depreciation per year, NVIDIA's accelerated product cycle is obsoleting old chip generations faster, not more slowly. This issue is increasingly important as Bury is correct in that each recent Nvidia chip cycle has been produced faster than the previous. And now, in the 2025-26 passage of the Big Beautiful Bill in its accelerated depreciation provision, financial need motivates companies to invest faster, reduce taxes, to reach profitability sooner. The new BBV depreciation rules allow front-loading of AI infrastructure CAPX depreciation in year one. This helps the financial profile of the AI

build-out as it allows front-loading depreciation, which does lower gap earnings. in early years, but it frees up cash flow by reducing taxable income. This is a super interesting discussion and yet very fluid in concept because mathematically in accounting, depreciation is linear. While in the real world of computing, more specifically, artificial intelligence, demand creation is growing exponentially. One doesn't have to look far to see what I'm talking about when industry insiders think about demand. At a November 6th meeting, a vice president at Google Cloud told employees that the company has to double its serving capacity every six months in order to meet demand for artificial intelligence services. Google raised its capital expenditures forecast

for the second time this year to a range of 91 to $93 billion following a significant increase to come in 2006. I don't have the answer to the question that Michael Burry and others are recently raised on depreciation scheduled and useful lives in video chips. I can only relay what those closer to the AI nucleus or relay domain. And chip useful life, those I've talked to say the industry is still getting function and value at a much older chips. Invidio technology that is five years old is not being used for new leading edge world models or deep neural networks. Instead, those chips and systems are being used for less data-intensive tasks that are closer to the end client. Think about general search or data retrieval queries that most consumers use regularly.

the Mag 7 accounting fraudulent, which kind of strikes me as an overly aggressive position. Nick Craig, CEO at the Earning Scout, and former head of earnings department at Sachs Research during dot-com time period, has expressed his opinion that while aggressive, the depreciation issue we're discussing is neither a crisis nor at all comparable to Burry's prior mortgage market security success. Investors, year-to-date, the Magus' raised capital spending by over 60% in 2025. Because of strong revenue growth, near to date, net operating cash flow grew by about 21.6%.

However, net free cash flow because of the CAPEX tsunami, the MAG 7's free cash flow turned negative in recent quarters. This negative free cash flow situation also occurred in 2022 when Meta took an unprecedented upfront spending on the Metaverse bets, just as TikTok and short form video format is picking up, team on the advertising revenue side, taking market share from previous social media leadership. Investors, many are worried that like Meta's ill-fated CapEx investments and bets on the Metaverse in 2022, most of which have yet to generate substantial revenue, Mag 7's current AI spending boom

will be money out the door with an indeterminate time and level of investment return. That their marginal return on invested capital, an oldest spending will be substantially less than their current asset light models, or worse yet, or a little or no extra revenue and take their marginal return on invested capital in the wrong direction over the coming few years. Lower, not higher. With all this set, investors and followers know that regardless of the outcome in this AI investment tsunami, our entire Oak Harvest team will be here for our clients doing what we can to plan for you and your family's future, regardless of what stage you're at in your career or in

your retirement. All content contained with an Oak Harvest podcast expresses the views of the speaker and is for informational purposes only. It is based on information believed to be reliable when created, but any cited data, indicators, statistics, or other sources are not guaranteed. The views and opinions expressed herein may change without notice. Strategies and ideas discussed may not be right for you, and nothing in this podcast should be considered as personalized investment, tax or leave. legal advice, or an offer or solicitation to buy or sell securities. Indexes such as the S&P 500 are not available for direct investment, and your investment results may differ when compared to an index.

Specific portfolio actions or strategies discussed will not apply to all client portfolios. Investing involves the risk of loss, and past performance is not indicative of future results.