SP500 Forward earnings estimate

The below is the facts 12 month forward EPS estimate, the earnings estimates are really bullish.

The challenge is how to be carry your concerns on AI capex boom on long term, and invest in the “NOW” where that investment book is driving excellent earnings growth.

1 Like

If you believe AI is a bubble likely to crash, which sectors are likely to prosper inspite of the crash. The market seems to think drugs and maybe energy. What else? Finance? Real estate? Agriculture?

AI is deeply embedded in the economy. It drives growth in many broader companies like Catterpiller, Cummins, Generac, Corning, Dell, etc, etc. An AI collapse could have long tentacles.

Actually beyond SW sector, AI has not produced any meaningful difference in productivity, or impacted margin.

This is all part of AI capex boom.. like DC build etc.

In relative terms or in absolute terms? I think the crash will impact many sectors. Financials/ Banks will be impacted because of lot of debt issuance, off-balance sheets are hidden in various places. Even if you think someone like Stryker, a medical company, if markets goes down, the wealth effect is going to have an impact on various elective surgeries, same goes for housing, real-estate, even if interest rates come down.

Except cash/ fixed income, it is difficult to predict which sectors will not be impacted.

I am struggling to find areas where you can hide.

2 Likes

I’m not sure fixed income will be spared. There’s a lot of debt financing in support of the AI buildout right now. Even if debt defaults are contained to just the riskiest AI related loans, if investors get spooked, fund managers could get forced to sell to meet redemptions. And when they get forced to sell, they’ll sell whatever is sellable. That may very well end up being the quality issues, which would knock their prices down.

There’s a reason market declines often end up with “the baby being thrown out with the bathwater”, and forced liquidation is often part of that reason.

Same, my friend. My bond ladder is already longer than I aim to keep it, and rates aren’t quite high enough for me to be comfortable extending it even further. My “high yield” savings account isn’t keeping up with inflation. And I’m having trouble finding clear value-priced stocks at the moment.

At this point, the thing that seems most sensible for me might be to increase my savings for appliance replacements. Both my air conditioner and my water heater are near the end of their lives. With the solar panels now in place, a high-efficiency heat pump replacement for each of them might be able to cut down my natural gas usage substantially while still keeping my annual electricity generation covered by the panels. (Current water heater is natural gas. Current furnace is natural gas.)

Ordinarily, I’d look for the solution with the lowest expected total cost of ownership over its anticipated lifetime. I am still inclined to think that way, but with high and volatile energy prices, tilting the math more towards lowest ongoing operating costs may not be such a bad idea…

Regards,

-Chuck

2 Likes

I am not looking at bond funds. I already have some exposure to $TLT, WFC.PR.L. and looking to add specific individual bonds, so distress sale scenario is not relevant here. I may mostly go with treasuries even if I end up getting lesser income. For me, liquidity, return of capital are more important than return on capital.

So… like the rest of us, you are waiting for interest rates to be more appetizing before additional capital commitment.

I am not waiting for higher interest rates, rather waiting for capital release. I already have treasuries, $TLT, $WFC.L, it is 12.5% of my portfolio.

In below Apollo’s Slok making the case that beyond tech sector AI has not shown any productivity gain, or margin impact.

Industries like Banking, financial, Healthcare, where AI adoption is high, and spending is high, are yet to see any tangible benefits. Citi is a top 10 company in adopting AI… and Jane, Citi CEO on AI

Second, our own adoption of AI is both earning and saving us money, tangibly improving productivity and efficiency through the bank. AI is accelerating our product development life cycle, we’re launching new products at lower cost and faster speed, typically seeing a 30% to 40% increase in the speed of product launches. Another example: 14,000 of our engineers use our main agentic coding tool, Devin, making it much more efficient to do work in-house. We’ve been in-sourcing work and have reduced our tech vendor spend by 30% year-over-year as a consequence, with a long way still to run.

Again, most of the gains are related to Citi’s ability to use AI for coding, and update their systems to support newer products. The earnings is lesser spend on developing/ updating new code… For the wider organization, implementing AI requires lots of process change. Right now, many management are putting brave face, partially to justify their AI spend, partially due to hopium.

Anyone who worked in corporate America knows, process re-engineering and change on any organization of over 200 folks is not that easy.

I am not sure about prosper part of it… consumer staples which rallied on AI scare.

Much of AI seems to be under construction. Most due in two years but many delayed due to backordered equipment or grid delays.

It’s maybe too early to see the productivity improvements in broad numbers. People do tell us computer coding is now much more productive due to AI. And we suspect that much writing and creativity benefits from AI input. Does that show in the numbers? Maybe.

I’ll just give my anecdotes. They match nicely with the Citi content. But there’s a catch.

Our teams have been pulling data into a common platform for the last 3.5 years. this is a HUGE enabler. Now that we have good, clean, curated data, we are building MUCH more quickly than before.

How AI helps:

I can directly (or the team lead) place the business requirements into a prompt and can iterate on that prompt to include .php or .net or .json, etc. code requirements.

The AI is providing an answer in less than 3 minutes for most request and no more than 10 for all requirements.

What this means:

I can iterate with operations USERS in real time. My teams can revise, resolve and enhance basic wireframe templates or prototypes while we are all in the same room or teams call.

No only does this enable the business person (Ops, supply chain, finance, maintenance, etc.) to understand their request better, it enables them to cut out several iteration cycles while dramatically reducing the error from intent along the way.

We have several internal and contract positions who have now been mandated to use Claude Code or some variant to cut the cycle time down.

We can do more, with less. The savings is about 80 percent with the depth in our teams no longer an issue and we are actively reducing our outside contract support.

The business is receiving their final product faster (by 75%) and with better long run parity to actual need. (vs iteration and sub optimal releases).

My forecast:

This will help the most backward industry accelerate their modernization MUCH more prevalently than to advance cutting edge businesses.

For new cases, they don’t have great data (short history, changing context, organic changes as the product burns in, etc.) and they are trying to align customer, product and development to value. Only the best/few will demonstrate maximum acceleration.

For old cases, the value is well defined, the business is tied to physical processes which have been in practice for decades or generations and their traditional investment in the space was quite low, with threshold to support new investments quite high (low margin businesses need to fund only the BEST use cases since they have limited capital and development capacity). Also, most of these businesses went on a CAPEX glut immediately after COVID supply/demand shocks.

A very large source of value is in traditional businesses. THEY will unlock value as a customer. The company working on the frontier model enhancements or supplying presence (neoclouds, etc.) will have to successfully capture rate in a highly competitive environment to win. Meanwhile, traditional brick and mortar operations will simply reduce tech overhead while dramatically accelerating their reduction of technology debt.

My budget got cut in half. I’m not so sure I won’t deliver the same value at a rate higher than pre AI. Again, with great data.

Equal weight S&P or Value indexes might be the play here.

1 Like

While some firms may enjoy slightly better margin, if margins can be gained by a business by adopting AI, the competitors will also do that. Competition will erase the margin or bring equilibrium. I am not expecting a sustained higher margin at the industry level.

1 Like

Some additional thought. The reason I highlighted XLP performance is, during Apr 2000 to Dec 2000, defensive stocks, XLP is a proxy for that, rallied, there was widespread rotation. Of course from 2001 to 2003 everyone declined.

Many argue, we are in 1997, and not in 2000. Only in hindsight you will know. Clearly financing concerns for AI has emerged.

1 Like