AI Is Now Carrying Wall Street's Profit Growth, and a 5% Treasury Yield Is Testing How Much It Can Hold
By Generational Wealth Investments | GenerationalWealth.biz
Stocks are not supposed to sit near record highs while the 10-year Treasury yield sits above 5%. Higher yields make every future dollar of profit worth less today, and for most of market history that math has pulled share prices down. This fall it hasn't, and the reason comes down to one source of profit: AI.
At Generational Wealth Investments, we don't chase hype, we decode the market. So here is the thesis, stated up front: the market isn't ignoring 5% yields. It is paying for them with AI profit growth. That trade works as long as the profits keep arriving, and it leaves the S&P 500 standing on a narrower base than the record highs suggest.
Key Takeaways
S&P 500 earnings are expected to rise about 30% from a year ago, with roughly two-thirds of the increase tied to technology and AI.
Semiconductor earnings are projected to surge about 136%, and 2 chipmakers account for more than one-third of the index's growth.
Profit growth has offset the valuation pressure from a 10-year Treasury yield above 5%, leaving little cushion if results fall short.
Taiwan Semiconductor reports Thursday, October 15, giving the first direct read on AI-chip demand this earnings season.
The Numbers: 30% Growth, With Two-Thirds From One Theme
Analysts expect S&P 500 earnings to rise about 30% from a year ago for the third quarter of 2026. The exact figure depends on whose data you use. Goldman Sachs and Bloomberg consensus sit near 27%, FactSet is closer to 29%. Every version tells the same story: this would be the 8th straight quarter of double-digit growth and the 3rd straight above 25%, a run that is rare outside of recoveries from deep recessions.
The more important number is where the growth comes from. Roughly two-thirds of the increase is expected to come from technology and AI heavyweights like Alphabet, Amazon, and Meta.
Run the arithmetic. Of roughly 30 percentage points of index-level growth, about 20 trace back to the AI trade. Everything else in the index, including banks, retailers, drugmakers, and industrials, supplies the remaining 10 or so.
One detail deserves attention. Alphabet and Meta are classified in the Communication Services sector, and Amazon sits in Consumer Discretionary. None of the 3 is officially a "technology" stock. Sector labels therefore understate how much of the index now depends on AI. An investor who feels diversified across 11 sectors may be holding the same bet in 3 or 4 of them.
The Engine Inside the Engine: Semiconductor Earnings Up 136%
Here's the part that matters most: semiconductor earnings are projected to surge about 136% from a year ago.
That figure is not a typo, and the mechanism behind it explains the whole market.
Step 1: Capital spending becomes chip revenue. The largest cloud and AI companies are building data centers at a pace with few precedents. Goldman Sachs projects their capital expenditures will rise 116% year over year in the third quarter, up from 87% growth in the second. A large share of every data-center dollar goes to chips, so one group's spending is the other group's revenue.
Step 2: Operating leverage turns revenue into outsized profit. Chipmakers carry heavy fixed costs in fabrication plants and research. Once those costs are covered, each additional chip sold is highly profitable. When revenue doubles and costs don't, profit more than doubles. Goldman notes that semiconductor gross margins are running near 70%, against a 15-year average of about 55%.
Step 3: A few companies move the entire index. According to Goldman, Micron and Nvidia alone are expected to supply more than one-third of the S&P 500's third-quarter earnings growth. That's 2 companies out of 500.
Put the chain together: a handful of buyers are spending record sums, a handful of sellers are booking that spending at record margins, and the result shows up as headline earnings growth for the whole market.
Why Stocks Have Held Near Records With the 10-Year Above 5%
A stock index's price is the product of 2 things: what companies earn, and the multiple investors will pay for those earnings. Treasury yields act on the multiple. When the risk-free rate rises, the present value of future profits falls, and investors pay less for each dollar of earnings.
The pressure on that multiple has been heavy. The 10-year yield crossed 5% in mid-September for the first time since 2023. On October 7 it reached 5.365%, its highest level since April 2002, with the 30-year near 5.73%. The Federal Reserve raised rates in September, and the minutes of that meeting kept another hike before year-end on the table.
Now look at what stocks did. On October 6, the S&P 500 closed above 7,800 for the first time. Yet LSEG data from late September showed the index trading at just under 19 times expected earnings, its lowest valuation since 2023.
Prices near records while valuations sit at multi-year lows can only mean one thing: earnings grew faster than prices. The multiple did compress, exactly as the textbook says it should when yields surge. Profit growth filled the gap.
There is a second-order point here that most coverage misses. At roughly 19 times earnings, the S&P 500's earnings yield is a little over 5%, about the same as a 10-year Treasury. The extra return investors normally demand for owning stocks over government bonds has shrunk to almost nothing. Nobody accepts stock risk for a bond-like yield unless they expect that yield to grow, which means expected growth is now doing all of the work. And two-thirds of the expected growth is AI.
The Bar Just Got Higher
This is why strong AI earnings are both the good news and the risk. If AI profits disappoint, the market loses one of its biggest supports, and it loses that support while the discount rate sits at a 24-year high. Earnings estimates would fall at the same moment the multiple has no cushion left. That is a double hit.
3 pressure points are worth understanding:
Margins. Goldman estimates that if semiconductor gross margins slid from roughly 70% back to their 55% long-run average, S&P 500 earnings could take a hit of about 10%. That scenario doesn't require AI demand to collapse. It only requires pricing power to normalize as supply catches up.
Expectations. Chip stocks are up more than 80% this year. A 136% earnings surge is already in the price, which means matching the forecast is not enough. Stocks priced for a blowout tend to trade on guidance, because markets care about the next 4 quarters more than the last one.
Financing. The data-center buildout is increasingly paid for with borrowed money, and that money now costs more than 5%. A higher cost of capital raises the return every new facility must earn to justify itself. The spending that generates chip profits is itself sensitive to interest rates, just with a lag.
There is also a quality question. Last quarter's reported S&P 500 growth came in near 50%, but iCapital estimates that close to 22 percentage points of it came from one-time items, including investment gains tied to AI holdings and tariff refunds. This quarter is a cleaner read on how much growth is repeatable.
The Next Test: Taiwan Semiconductor Reports Thursday
The first hard data point arrives Thursday, October 15, when Taiwan Semiconductor Manufacturing Company (TSMC) reports third-quarter earnings. The conference call is scheduled for 2:00 a.m. Eastern, so U.S. investors will wake up to the results.
TSMC matters because it manufactures the advanced chips that Nvidia, AMD, Broadcom, and Apple design. It sees AI-chip demand before almost anyone else, which makes its report the closest thing to a direct measurement of the cycle.
The company guided third-quarter revenue to between $44.6 billion and $45.8 billion, up from $40.2 billion in the second quarter. The midpoint implies roughly 12% growth in a single quarter. It guided gross margin to between 65% and 67%.
4 things to watch:
Whether revenue lands at the top of the guided range or above it
Whether gross margin holds near the high end, since margin is where the 136% story is most exposed
Fourth-quarter guidance and any change to capital spending plans
Management's read on whether AI demand still exceeds what TSMC can supply
3 Scenarios for the Weeks Ahead
Scenario 1: Beat and raise. TSMC tops its range, holds margins, and guides higher. The AI profit engine is confirmed for another quarter, and earnings keep absorbing yield pressure. Stocks can hold near records even if the 10-year stays above 5%. The cost is that concentration deepens and the bar rises again for Nvidia and the hyperscalers later in the season.
Scenario 2: In line, with softer margins. Revenue lands inside the range, but gross margin drifts toward the low end or the outlook turns cautious. Nothing breaks, but an 80% year-to-date rally in chip stocks has little room for "fine." Expect leadership to wobble and attention to shift toward whether the other 490-plus companies can carry more of the load.
Scenario 3: Guidance disappoints. A weaker outlook from the company at the center of the supply chain would call the 136% projection into question. With yields at multi-decade highs, there is no valuation cushion to absorb lower estimates. This is the scenario where earnings and multiples fall together.
What Would Prove This Thesis Wrong
A thesis is only useful if it can fail. Here is what would tell us AI earnings are not the main thing holding this market up:
AI results disappoint and the index holds anyway. That would mean breadth is real. iCapital notes that expected earnings strength is spreading beyond technology, with the share of sectors showing positive growth near post-pandemic highs. If that broadening carries the index through a chip stumble, the base is wider than we think.
AI companies beat and raise, and stocks fall anyway. That would mean the discount rate has taken over and earnings are no longer enough to offset yields.
The 10-year drops back below 5% and leadership rotates away from AI without index-level damage. That would mean the market was leaning on rates relief, not chip profits.
What This Means for Your Portfolio
The takeaway isn't that AI is a bubble or that the rally is doomed. The earnings are real, and that is exactly why stocks have held up against the highest long-term yields since 2002. The takeaway is that a record-high index and a diversified index are not the same thing.
A few questions worth sitting with:
How much of your portfolio is the same AI bet under different sector labels?
If semiconductor margins normalized, which of your holdings would feel it first?
Are you watching guidance, or just the headline beat?
Thinking these through before Thursday is a better use of your million-dollar hours than reacting to a 2:00 a.m. headline.
Stay Ahead of the Market Every Day
This is the Generational Wealth Community, your pathway from knowledge to legacy. We decode what moves markets. Follow or subscribe for the market setup every morning, and join the community at https://www.generationalwealth.biz/community to tell us what you're watching.
⚠️ Educational Disclaimer: This content is produced by Generational Wealth Investments for educational and informational purposes only. Nothing here constitutes financial or investment advice. Markets are volatile. Always do your own research and consult a licensed financial professional before making investment decisions.
