A Fund Manager Says the Old Market Indicators Broke in November 2022. His Firm Just Launched ETFs for the New Ones
A fund manager claims the economic gauges investors have trusted for decades stopped working the moment ChatGPT launched, and now his firm is selling ETFs built around what replaced them. Before you buy the theme, check who wrote it.
This post may contain links from our sponsors and affiliates, and Flywheel Publishing may receive compensation for actions taken through them.
Harbor Capital Advisors has a big claim and a product to go with it. Kristof Gleich, the firm’s president and chief investment officer, says the indicators investors have long used to read the economy stopped tracking the stock market after OpenAI released ChatGPT in November 2022. His firm is now selling ETFs built for the economy he says came next.
Harbor’s argument is real. So is the commercial interest behind it. If you’re near retirement and keep hearing that AI changes everything, consider both.
Harbor’s newest fund is the 800VDC AI Datacenter Ecosystem ETF (NYSEARCA:HUMM). Its prospectus lists a net expense ratio of 0.49%.
Harbor Says Investors Are Reading the Economy With the Wrong Gauges
Gleich made the case at a press briefing. He told the room the AI economy “requires us to look at different frameworks, different benchmarks, new units of measure, and frankly you need a different language to describe, measure and invest in it.”
His evidence was the manufacturing purchasing managers’ index (PMI), a survey showing whether factory activity is growing or shrinking. Gleich pointed to its longstanding relationship with the S&P 500. He said the two moved in together until ChatGPT arrived. After November 2022, the factory indicators stayed roughly flat while the market rose sharply. 24/7 Wall St. could not independently verify that divergence, so read it as Harbor’s argument, according to Harbor Capital Advisors.
A Two-Phase Boom With an Inflation Catch
Spenser Lerner, Harbor’s head of multi-asset solutions, divides the boom into two phases. First comes the build-out through decade’s end. Lerner says hyperscalers and data center owners are investing close to $1 trillion into AI infrastructure this year, or about 3% of GDP, with roughly $1.3 trillion in 2027.
Lerner says this spending accounts for 35% to 50% of current U.S. GDP growth of 2.5%. The cost: demand strains power, chips and specialized labor, pushing up inflation and real interest rates. All of that expansion has to be powered, cooled and connected by somebody, which is why we pulled together seven companies doing exactly that in a free guide to the AI infrastructure trade.
The second phase is adoption, when agentic workflows spread from big tech to small businesses. Lerner expects that to cut production costs, widen margins and ease inflation while growth picks up. He called it a “Goldilocks combination.”
He also described a scenario where faster growth shrinks the debt burden, requiring 3.5% to 4% growth, according to Harbor Capital Advisors. He called that “a more optimistic outlook” held by some economists, and said it sits outside his base case.
Harbor’s Equity Chief Says AI Isn’t a Bubble. Here’s What Has to Hold
When asked whether AI stocks are in a bubble, Justin Menne, Harbor’s head of global equities, said simply, “No.” He cited a combined backlog of roughly $2.4 trillion at the end of the second quarter at Oracle (NYSE:ORCL | ORCL Price Prediction), Microsoft (NASDAQ:MSFT), Amazon (NASDAQ:AMZN) and Alphabet‘s (NASDAQ:GOOGL) Google.
Menne’s model of a one-gigawatt data center showed operating margins of 50% to 60%. With leverage, return on equity was approaching 40%, according to Harbor Capital Advisors.
Caveats: backlogs are future revenue not yet earned and contracts can slip. A model of a typical facility is still just a model. And the person saying there’s no bubble runs equities at a firm launching AI funds. For Menne to be right, those contracts must turn into revenue on schedule and real data centers must earn margins close to the projected ones.
Check What These AI Funds Own Before Buying the Name
Harbor launched ETFs in August benchmarked to five leading AI labs, including Anthropic, OpenAI, xAI and Google DeepMind, according to Financial Advisor magazine. Most are privately held. A benchmark is the standard a fund measures itself against. Read each fund’s holdings and structure before assuming you’re getting shares of these labs.
HUMM had traded for just 11 sessions as of October 1. That gives you very little history to judge it by, and no asset base or holdings figures were available to review.
Buying a Theme From the Firm That Wrote It Calls for Extra Proof
A thematic fund from the firm that defined the theme calls for extra examination. My view: the fund still has to prove itself, given its short trading history.
Here’s a concrete test. Look at the holdings. Compare the expense ratio with a broad index fund. Check the fund’s age and asset base. Then ask whether the thesis depends on a specific technology arriving on time. HUMM is tied to an architecture Harbor expects in 2027, and that schedule is a timing risk a broad index fund doesn’t carry, according to Harbor Capital Advisors.
Contact [email protected] for any questions or corrections.
Jake has been been working in financial media for almost 15 years. He focuses on all things personal finance for 24/7 Wall St. with high hopes to educate and entertain. Most recently, Jake spent 12 years working various roles at The Motley Fool. He started copy editing fool.com content, worked on premium and marketing campaigns, and helped launch The Ascent, a personal finance brand.
His work has been featured on platforms like MSN, Yahoo Finance, USA Today, and more. He's written about credit cards, social security, ETFs, savings accounts, and just about anything else you can imagine when thinking about money. Jake love to cook, play golf, and tell people he's never had a cavity. (It's true!)
