By Ellen Hazen, CFA®, Chief Market Strategist
What We’re Watching in June
• Federal Open Market Committee (“Fed”) meeting June 16-17. This will be the first meeting with Kevin Warsh as chair. Markets are expecting no action on the overnight federal funds rate (although are pricing in a rate increase later in 2026), but Chair Warsh may announce any number of changes to the Fed’s process, including shrinking the balance sheet (currently $6.5 trillion), eliminating the so-called “Dot Plot” in the Fed’s Summary of Economic Projections (SEP), a change to the 2% inflation target, and removing the maximum employment part of the dual mandate.
• SpaceX IPO. SpaceX is raising $75 billion in an initial public offering with a target valuation of $1.75 trillion to $2.0 trillion. (The stock will not be part of the S&P 500 Index for at least 6 months, due to listing requirements, but will become a meaningful weight in the Nasdaq 100 immediately). The SpaceX S-1 filing lays out an ambitious plan to integrate xAI’s large language model with StarLink’s satellite network and xAI’s Colossus datacenter, with IPO proceeds to be used to fund continued cash needs for satellite launches and replacements and a new semiconductor facility called Terafab. In the most recent 12 months, the company reported a $5 billion loss and spent an additional $20 billion on capital expenditures
• Labor Market Continued Strength
o The May jobs report was unexpectedly strong, with 172,000 new jobs created in May, compared to analysts’ estimates of only 88,000. Average hourly earnings increased by 0.3% month-over-month or 3.4% year-over-year. June jobs will be announced on July 2
o Will we see more layoff announcements? A number of companies (Block, Dow, HP, Meta, Amazon) have announced layoffs at least partially justified by AI. However, monthly average private sector job layoffs in total have remained relatively stable at 75,000 jobs per month over the past year, suggesting that this is more of a catchy narrative than a true trend underpinned by data
• Inflation to be reported June 14 (CPI) and June 30 (PCE). Both measures of inflation have remained stubbornly above the Fed’s 2% target for over 5 years, which has served as the primary reason the Fed has not cut interest rates further
• Tech stock performance – can it continue? The S&P 500 Information Technology sector index has appreciated by 17%, more than double the S&P 500 Index’s return of 8.4%
• We expect interest rates across the curve to remain above 2025 levels. Most maturities have increased by about 0.40% year to date, as inflation has remained higher than expected
• Continued increases in hyperscaler capital spending. Analysts have increased their estimates for the four large hyperscaler companies’ 2026 capital spending (Meta, Alphabet, Amazon, and Microsoft) from $467 billion as of the end of 2025 to $712 billion today. We expect this to continue to increase, particularly given Alphabet’s recent $80 billion equity raise
AI and the Return of the Productivity Paradox: Individual Adoption vs. Organizational Redesign
Regular readers will recall that we closely follow hyperscaler capital spending, which has increased to over $700 billion in 2026 from just the four largest among them. Is this money going to be well spent, generating a return? Or will it prove to be wasted? One way to assess whether this is generating value is to examine the productivity data published by the Bureau of Labor Statistics (BLS). Alas, as Robert Solow famously said in 1987, “You can see the computer age everywhere but in the productivity statistics.” The most recent BLS data show a modest increase in labor productivity, up 0.3% quarter over quarter and up 2.8% year over year, hardly a productivity boom and still below the 3.5% average seen in the years 1995-2005 during the internet buildout.
Innovations can take years before increasing productivity. This is not entirely surprising. Two things can be true at once: first, that AI will prove to be a productivity enhancer, and second, that it will take time to show up in the data. Most major technological innovations took time to generate economy-wide productivity, from electrification to railroads to the internet. More specifically, these innovations took years or even decades to diffuse throughout the economy before making a substantial impact. Paul David’s seminal 1990 paper highlighted that while Thomas Edison had introduced incandescent lamps in 1879, and both New York and London had central electricity generation by 1882, twenty years later only 3% of residences used electric lighting. It wasn’t until the 1920s – when a full 50% of factories’ power needs were provided by electricity – that businesses redesigned their workflows around electricity and worker productivity soared. Steam engines similarly took decades to diffuse throughout the economy. And although the first computer (ENIAC) was invented in the 1940s, the internet revolution lagged even the IBM PC 1981 introduction by over a decade. We believe that AI will likely follow the same pattern: slow diffusion throughout the economy, followed by a productivity payoff once diffusion is sufficiently broad, people have been trained, and companies have overhauled their business workflows.
AI adoption is real – and faster than PC adoption. We know that AI is being adopted. The US Census Bureau’s Business Trends and Outlook Survey shows a steady increase in corporate AI usage over the past six months, from 21% to 23% – and 37% among larger businesses with at least 250 employees.
The St. Louis Fed also tracks adoption – both personal and at work – and shows current adoption at 58%. Interestingly, the St. Louis Fed data says that non-work usage is trending about 10 points higher than work usage. And (also from the St. Louis Fed), it is being adopted more quickly than the PC or the internet.
Today’s AI adoption seems to be more driven by individuals than employers. Individual usage creates local efficiency, while measurable productivity requires firm-level redesign. This is likely why $700 billion of capital spending (investment) hasn’t yet translated into output gains. Without business redesign, ad hoc worker AI usage will continue to provide limited productivity boosts. This is consistent with academic literature. MIT economics professor (and Nobel laureate) Daron Acemoglu estimates that AI will increase productivity by only about 0.5% per year. This may be true if applied only to individual tasks, but the bigger opportunity is when AI is applied to redesigned workflows.
Productivity often falters before it increases. We should expect continued evidence of the “J-curve” – investment spending in new general-purpose technologies precedes payoffs, because business models, human capital, and processes need to evolve to take full advantage of the new technology. J-Curve coiner Erik Brynjolfsson found that productivity actually slows down at first, increasing only when all the requisite conditions are met.
The idea that we may be in the low part of the J-curve is supported by the data so far. As previously mentioned, labor productivity has not yet materially increased. Similarly, corporate layoffs have been stable (despite the impression that media headlines might give), again suggesting that companies have not yet wholesale substituted AI for labor costs.
A few individual productivity enhancing examples, but not enough to move the needle. When searching for examples of individual businesses that have generated a positive return on AI investment, a few examples pop up, validating the possible improvements at the micro, individual project level. According to McKinsey, UK insurance company Aviva redesigned its claims process to be centered around AI and not only saved costs and shortened claims resolution time but increased its net promoter score by 7x. Unilever also claims cost savings from having reimagined its supply chain workflow. At the same time, these (perhaps isolated) examples are not driving corporate-wide improvements: earnings estimates for both Aviva and Unilever have declined in recent months.
So how will we know when companies are starting to ascend the right-hand side of the J-curve? We will see it in corporate earnings increases, and it in broader productivity statistics. We will hear stories of total business reorganizations. It may be that new, AI-native businesses will be more easily able to exploit the advantages that AI brings than legacy companies are, just as cloud-native companies outperformed legacy internet companies.
AI will likely drive a productivity boom, but requires broad technological diffusion, organizational change, broad adoption, and human capital investment as well as simple AI tool availability, usage, and narrow adoption. Both the timing and magnitude remain uncertain.
From an investment perspective, this gives us optimism for continued weighting to risk assets like equities, as an individual company’s equity performance has historically been associated with its productivity (measured by operating leverage). We expect that over time, AI-driven returns will broaden beyond just the technology companies enabling AI today, to tomorrow’s corporate AI users. Higher productivity leads to higher margins, which should lead to a broader earnings benefit and broader equity participation.
Looking not just at individual company performance but at broad asset classes, Citigroup identified several previous “Goldilocks scenarios” when the US had both strong economic growth and muted inflation (as one would expect in a productivity boom) – in these cases, equity markets rose and real interest rates remained higher for longer. Consequently, if AI productivity plays out similar to other technological innovations, accelerating growth and dampening inflation, equities can continue to perform well even as interest rates may be higher. The benefit of AI adoption is not yet showing up in the official productivity data, which is to be expected given current penetration levels – but when it does, we believe the economy and equities will be beneficiaries.