Amer­icas CIO View

The performance rotation YTD comes without an EPS growth leadership change

Last year, we discussed diversifying from US large Growth stocks and the Magnificent 7. Heading into 2026, we suggested that broadening US equity market performance should be expected in early 2026 and seen as healthy, but that a rotation away from large US Growth stocks and the Magnificent 7 or 10 should only last for a relatively short time and limited in magnitude, assuming the economic expansion is long-lived as we expect. Please refer to our December 22, 2025 note, “Twas the night before a leadership change?”

Here at the start of March, Russell 1000 Growth delivered a -5% year-to-date (YTD) and Russell 1000 Value delivered 7% return. Among the biggest short period spreads since early 2022. While the S&P 500 is up just slightly YTD, Magnificent 7 is down 5%. At the opposite ends of the sector spectrum, Tech is down 6% and Energy is up 25% YTD. This is a big rotation beyond a broadening as seen with S&P 500 equal-weight 2-3% ahead of the weighted index and small caps up 6-8%. Yet, as we finish 4Q25 earnings season, we continue to see Tech and the Magnificent 7 deliver 25-30% y/y EPS growth with upward revisions to 2026. With the rest of S&P EPS ex. Energy and Financials deliver 6%; up 4% at Energy and Financials up 14% y/y. The market is certainly forward looking on earnings, but estimate trends stayed strongest for the Magnificent, Financials, Health Care, Utilities and yet wobbly elsewhere.

Rotation out of Tech and within Tech: The rotation within Tech deserves scrutiny

We currently think about the Tech sector and a few related titans outside it as divisible into three pieces: 1) Capital expenditure (Capex) Spenders – hyperscalers and artificial intelligence  (AI) model/service leaders, 2) Capex Recipients – AI semiconductors, memory and other tech hardware & equipment, 3) Capital Light – software and other tech based service providers mostly to enterprises that live on specialization, innovation and other intangibles like reliability, user networks and standard setting. The rotation within Tech has seen group 3 plunge while group 2 soars. Software & Tech Services is down 20% YTD while Tech Hardware including Semiconductors is up 19%. Group 1 has been flattish and mostly in line with the S&P 500, mostly a consolidation.

The consensus view is essentially that capex at group 1 is too high for attractive returns on capital to be earned anytime soon and yet, somewhat contradictory, confidence is rising in the capabilities of the AI being developed, calling into question the profitability/longevity of businesses in group 3. We understand these concerns, but we think the situation calls for greater scrutiny of which hyperscalers are merely cloud hosting vs. AI service providers. We see value in hyperscalers with proprietary leading AI abilities, lots of vertically integrated uses, leadership in datacenter hardware configuration and operating efficiency. Group 2 is booming, recent guidance and outlooks were very strong, yet if group 1 can’t earn decent returns on this investment within a credibly visible time then capex will slow. We see value in group 2 for now, but these are cyclical and highly competitive businesses.

Will AI eat all of software? Will it eat all our electricity? Will it eat all of humanity?

In our US datacenter capex and Return on Invested Capital (ROIC) model, we raised capex forecasts again since the version published in our November 25, 2025 note, “AI is growing quickly and is very hungry: S&P 500 provides the essentials.” Our model forecasts roughly $3.8trn of gross capex from 2024 to 2030 on new US datacenters (chips and everything inside and onsite) and about 3% US electricity demand growth through 2030 from these datacenters. Our model suggests that for ROIC to equal a reasonable cost of capital (WACC): 1) chip/compute costs need to moderate, 2) electricity prices must not surge, and 3) $750bn of revenue must be generated by these datacenters and the AI services supported by 2030.

We and many are contemplating $750bn to well above $1trn of revenue stemming from this capex in 2030. This would be 2-3% of US gross domestic product (GDP) in 2030. This revenue could take share from existing businesses and labor or it could lead to greater output than most GDP forecasts include. We doubt that all of this revenue will be extracted from labor share. A guess would be that half comes from less workers, yet history suggests none. AI driven productivity might accelerate kept worker productivity growth by 0.4% annually later this decade. As always, moving displaced workers to new and better uses will be key.

DWS CIO Day, March 2027 Strategic Targets:  S&P 500 target stays 7500

We set targets before the strikes on Iran. Yet, we see our 12-month targets across asset classes as appropriate assuming conflict remains contained. We think seeking alternatives to US equities in pursuit of further diversification is sound, including Japan, Europe and select Emerging Markets (EM) equities, rather than reaching deeper into non-Financial US value equities or small caps. We also advocate active strategies and modified S&P 500 index constructions.