Episodes

Friday Feb 14, 2025
Friday Feb 14, 2025
How much electricity will AI need?
To train AI models companies use graphics processing units, also known as GPUs. They are now starting to build larger clusters of GPUs, which requires even more electricity. How much electricity you may ask? AI data centers use about 30 Megawatts of electricity at a time. If you don’t understand megawatts, let’s just say it’s a lot of power. Picture 30 Walmart stores and how much electricity they use at any given time, that is estimated at 30 megawatts. Fast forward five years into the future when there will be more data centers and larger AI models. It is estimated they will require 5 gigawatts of electricity. 5 gigawatts is a huge amount of energy, it is about the same amount of energy needed to power a city like Manhattan in New York. Also, a big concern is within the next five years these massive data centers could consume up to 17% of US electricity. You may be thinking just build more power plants. The problem is data centers can be completed within 18 to 24 months, but to build a power plant can take over three years and that’s provided all permits and regulations are met on time. There’s also the concern of how do you get that energy to the data centers, you’re going to need more transmission lines, but that can take 10 years or longer to get that task completed. Wind and solar are not the answer because data centers need power 24 hours seven days a week and when the sun goes down or the wind stops, there’s no power. I see some roadblocks ahead with fast moving AI, maybe we need to slow down a little bit?
Mag Seven capital expenditures could be a big problem!
The Mag Seven, which includes Apple, Alphabet, Amazon, Microsoft, Meta, Nvidia, and Tesla has been a group that has dominated the stock market the last couple of years. Much of the excitement around the stocks have been tied to advancements in AI, but there has still been little evidence these companies (outside of Nvidia) have been able to profit from the trend. A major concern I have is these companies are investing tons of money and the big question is how profitable will these investments be? It is estimated Alphabet, Microsoft, and Meta will spend $200 billion on artificial intelligence this year alone and their budgets have continued to grow. If we look at total capital expenditures, also known as capex, the budgets have grown immensely for many of these companies. Amazon is projected to spend around $105 billion on capital expenditures in 2025, up 27% from 2024, which came after a 57% increase over 2023. Microsoft has guided to $80 billion in capex for its fiscal 2025, up 80% from 2024, which was up 58% from the year before. Alphabet estimated capex of $75 billion in 2025, up 43% from 2023, which was up 63% from 2022. Meta has a forecast of $60 billion to $65 billion of capex in 2025, up 68% at the midpoint from 2024, which was up by 37% from the year before. The big problem with major capex is investors won’t see much of a difference in earnings, but there will be major hits to free cashflow. Capex is generally expensed or depreciated over time, which means it won’t hit earnings in a major way initially, but it could weigh on earnings growth over time as that expense remains for years to come and potentially grows if capex budgets continue to climb. As an example, Meta is projected to see $68 billion of net income this year, but free cash flow could slide 25% to $40 billion. Investments of this magnitude need to pay off, especially considering the high valuations for these stocks. Time will tell if these investments work out for all these companies, but I must say I’m skeptical they will all be winners from this movement 5-10 years from now. Investors need to look at the full picture and understand all the moving parts, which includes how all the financial statements work together. At our firm we don’t just look at earnings, we also want to see good cash flow and a strong balance sheet.
Producer Prices come in hotter than expected
The Producer Price Index, also known as the PPI, showed prices in January climbed 0.4% compared to last month. This topped the expectation of 0.3% and led to an annual increase of 3.5%. Core PPI, which excludes food and energy, produced an annual gain of 3.4%, which was lower than last month’s reading of 3.5%. While these data points were a little hotter than expected, economists now have inputs to estimate the closely followed PCE report. It is interesting that after the release of the CPI and PPI, which were both higher than expected, estimates for core PCE actually look quite favorable. On a monthly basis core PCE is expected to show a 0.22% increase, which would be a nice deceleration from December’s reading of 0.45% and on annual basis estimates are looking for a reasonable 2.5% increase. We will have to see what the actual results look like for the PCE later this month, but with these reports now in hand I continue to believe that while inflation is not at the Fed target, I still don’t see it as a major problem.
The True Cost of Credit Card Interest
Everyone knows that paying credit card interest is a bad thing, but it’s less well known how that interest is accrued. Interest is calculated using an average daily balance method, which means every single purchase begins accruing interest immediately. Purchases made on a credit card throughout a monthly statement period increase the outstanding balance. After a month of spending, if the full statement balance is paid by the due date, which is generally 20 to 25 days after the statement period ends, no interest will be due, even though it was accruing during that time. This is known as the grace period which is essentially an interest-free loan on those purchases. For example, if you spend a total of $5,000 through a statement period during January, you will not need to make a $5,000 payment until the end of February to avoid any interest. However, if you do not make that full payment by the due date in February, your grace period is void and you will owe accrued interest from the date those purchases were made in January, not from the due date in February. Also, any additional purchases made in February and afterward begin accruing interest immediately without a grace period, even though those statement periods have not ended yet. Since interest is calculated using the average daily balance method, the unpaid balance and interest compounds on itself making it more and more difficult to pay off. Credit cards have a monthly minimum payment, which is usually $25 to $50 dollars, which paying prevents a mark on your credit report, but it does not stop interest from accruing. Credit cards can be a great tool as they can give you points and fraud protection, but those benefits are greatly outweighed when a balance is being carried.
Companies Discussed: Dollar Tree, Inc. (DLTR), Grand Canyon Education, Inc. (LOPE), Mondelez International, Inc. (MDLZ) & Phillips 66 (PSX)
Comments (0)
To leave or reply to comments, please download free Podbean or
No Comments
To leave or reply to comments,
please download free Podbean App.