As of today, all MAG7 stocks are experiencing a negative year. However, they aren't all falling for the same reason. They aren't facing the same stress, and they won't all emerge with the same results. The market is currently asking MAG7: when and how will the capital you've invested turn into profit? Even asking this question is, in my opinion, a sign of a turning point. Nobody was asking this in 2023. They didn't ask it in 2024, and even partially in 2025. Growth figures were accepted, and stocks were rising accordingly. Now, all these assumptions are being questioned, and this questioning isn't coincidental; it's structural. I see four reasons for this. What I'm about to say isn't to say don't invest in MAG7; on the contrary, I wanted to explain that the US stock market won't move without MAG7, and what position MAG7 is in in the long term. I'll share 2-3 posts about this; I'll share what's happening today and what might happen tomorrow. It's a bit long, but there's nothing I can do about it.
The first reason is the scale that CapEx has reached. The five largest hyperscalers' total CapEx by 2026 is roughly $700 billion, with about three-quarters of that going directly to AI infrastructure. In the pre-AI era, CapEx as a percentage of operating cash flow was around 25%. Now it's pushing towards 70%. This shift is dividing institutional and individual investors. One side sees it as a necessary infrastructure investment. The other is waiting for it to prove its return. And the lengthening of this waiting period continues to put pressure on their stocks. I'm on the first side.
The second reason is the divergence itself. These seven companies don't share the same problem. Microsoft's problem is that Azure growth isn't currently justifying the CapEx burden. They spent $37.5 billion in a single quarter, over $100 billion annually, but Azure has fallen short of expectations for two consecutive quarters, and Copilot penetration hasn't progressed at the expected pace. The market responded clearly and harshly, and the stock fell 10% in a single session after the earnings report. This isn't a reaction; it's a message.
$AMZN | Amazon has a different picture, but the same problem. They spent $128 billion in 2025, with a plan for $200 billion in 2026 – a 56% increase in a single year, the largest single-year capital expenditure planned in corporate history. Simultaneously, they're laying off 30,000 people. This approach clearly shows they're financing machines, not people. AWS growth has regained momentum and the backlog is at record levels, but the $200 billion cash outflow is severely squeezing free cash flow. If Amazon plays this game right, it will be the biggest winner in this cycle. If it plays it wrong, investors like me will have to wait a long time.
$TSLA | Tesla, on the other hand, stands in a completely different category. The problem for the rest is the rate at which CapEx (Capital Expenditure) is converting to revenue; Tesla's problem is the revenue base itself that will support CapEx. EV sales have declined for two consecutive years, Chinese competition has squeezed margins, yet spending is planned for 2026 exceeding $20 billion, with the vast majority of that money flowing into a platform that isn't yet generating revenue: Optimus, Cybercab, and AI infrastructure. The stock's high price is based on faith in this platform, but that faith is directly linked to the growth of the EV base. If the EV base doesn't grow, the scaling of the Robotaxi model is hampered.
$AAPL | Apple's position is unclear. While others are spending a total of $650 billion, Apple is spending a small fraction of that, partnering with Google Gemini for Siri and pursuing a strategy that assumes AI models will become commodities. This strategy only works when others are struggling. Once the market stabilizes, Apple's own problems come to the forefront because the Siri revision has been delayed and its AI product vision remains unclear. If AI ceases to be a commodity and becomes a sustainable differentiating tool, the relative advantage gained from lower spending will turn into a permanent competitive disadvantage, and 2026 will be the year testing the limits of this strategy. This is the third reason for Apple investors, and it's the most critical one: why the two names structurally diverge in this picture.
$META announced a CapEx plan of $115-135 billion for 2026, the stock fell sharply, then quickly recovered with strong revenue guidance. The reason for this recovery is important because the market questions Meta's spending less than others, it can see the conversion to revenue. The integration of AI into the advertising algorithm drives ARPU, the Llama open-source strategy both grows the developer ecosystem and strengthens competitive defense, and agent workflows could take advertising to a new dimension.
The picture diverges similarly at $GOOGL | Alphabet. Gemini has gained traction in both the enterprise and consumer segments; Apple's choice of Gemini for Siri further illustrates this; YouTube is surpassing major media companies in advertising revenue; and TPU chips are providing both cost advantages and a new revenue stream. Alphabet is the only company that successfully manages the search, cloud, AI, and advertising cycle simultaneously. In my opinion, Meta and Alphabet are the two companies within this group that most clearly link AI spending to revenue.
$NVDA | Nvidia stands apart because, regardless of how much the rest of the Mag7 spends, the vast majority of that money flows to Nvidia. The YTD negativity is a result of incredibly high expectations, not fundamental deterioration.
The fourth reason is a structural break that seems to go beyond all this company-specific analysis. While Mag7 is falling, the rest of the index is rising. The cap-weighted S&P 500 has been almost flat year-to-date, with the average stock in the index up about 7%. The equally weighted index has outperformed the Mag7 ETF by about 14 points this year. Mag7's profit growth is expected to remain at around 18% in 2026, the slowest pace since 2022. The gap with the rest of the index is closing rapidly, and as the gap closes, capital is moving.
It's important to consider what this movement means. We've moved from a world where passive fund flows were directed to these seven names. These names make up about a third of the S&P 500. But they're no longer dragging the index; they're weighting it. Historically, after periods of concentration, the broader market has emerged stronger from this transition, just like the currently weighted S&P 500. And as this transition strengthens, I think investment in AI will return to these seven names, the real winners. The reason is this: MAG-7 created this wave, financed this wave, and is now settling the score. As they turn their spending into profit, investors will turn their attention back to MAG-7. I believe we will gradually see this in the second half of 2026.
