US labor's share of total income at a series low

Higher energy prices providing most of the boost to recent increase in EA inflation

Despite resilient economy, US consumers are not very happy

Weakness in Chinese investment is spreading

Benchmark revision shows less worrisome US inflation

Recent increases in US long-term rates surprisingly little impacted by inflation expectations

Even with strong capex, US nonfinancial corporations are generating more cash than they are investing

Investor sentiment worldwide is positive, especially in AI-led Asia

Sectoral shift in US job openings

Meteoric rise in AI-related semiconductor exports fueling South Korea's export boom

Regional Fed surveys showing further manufacturing acceleration in August

Sharp drop in China demand for oil helping to limit impact of Middle East conflict on the price

Decades high real price of copper posing physical challenge to investment boom

Rising US long-term real yields point to strong demand for investment funding

US retail sales accelerate even as consumers remain gloomy

Retail gasoline price still well above pre-conflict level and now rising again

PMIs showing widespread revival of global economy

Crude oil price has returned to pre-conflict level but traffic through Hormuz still subpar

Surging orders for computers and electronics propelling durable goods orders

Next year's inflation forecasts are on the rise and above target

Disrupted global supply chains placing upward pressure on US goods prices

Construction of data centers has overtaken transportation construction in US

Elevated energy prices putting intense upward pressure on input prices around the globe

US is big leader in global surge in private AI investment

Larger-than-usual income tax refunds this year are supporting consumer spending

Our global data offering is unparalleled in terms of coverage, quality, and organization. We offer 290+ databases from 2500 government and private sources, a wealth of key survey and forecast data covering the world’s economies, and a robust offering of archive databases that drive machine learning for big data applications.

Optimized for precision analytics, our data is seamlessly integrated with Microsoft Office and all leading statistical packages through our DLX® software and platform independent HaverView™ website. We further offer direct database feeds, client server or cloud-based delivery, and RESTful API.


Privacy Policy

Disclaimer

Contact us

Service Health

A premier time series data platform for the global strategy, research, and quant communities. 

New York

Tokyo

London

Singapore

Economy in Brief

Viewpoints

Contact Us

Our Data

Our Products

What We Do

Our Authors

About Us

Data News

Featured Data Additions

Resources

Release Calendars

Training Center

Webinars

Newsletters

Download Software

For Clients

Live Updates

Mickey D. Levy

Before the midterm elections, the buildout of AI data centers received a lot of attention and was identified as being a significant factor contributing to strong economic growth; at the same time, there were mounting worries of overinvestment and related excess indebtedness. Post midterm elections, significant discussion has turned to the downside: if newly-elected state legislatures and regulators clamp down on the AI data center buildout, commentators now fear there will be significant damage to the economy.

My assessment is the AI buildout has added to the economy--although its net impact on GDP has been relatively small--and a clampdown on the AI data center buildout would dent intermediate-term economic performance, but have a relatively modest impact on GDP growth. There are reasons to be optimistic. Here's some observations to support this view.

Consumption has consistently been the biggest contributor to economic growth. Consumption comprises roughly 67% of GDP, and in the last several years, its growth has contributed approximately 1.8 percentage points to the average 2.7% annualized growth in real GDP.  Business fixed investment has also been strong, driven by the boom in AI data center build outs (Chart 1). The BFI share of GDP has risen to 17%, its highest since the early 1980s, and 10% of business investment is attributable to investment in AI. That means AI investment is roughly 1.7% of GDP.

261111ML1.jpg

But over 70% of all of the capital spending in the AI data center buildout--including semiconductors, telecommunications equipment, and computers/peripherals--is imported (Chart 2) (Sources: World Trade Organization Global Review 2026; Jason Furman, "The AI Boom's Two Engines", the Peterson Institute for International Economics", November 2026).  Those capital goods are produced overseas and as such, the largest portion of the investment in AI data centers is not domestic production and is not measured in GDP.  No question, the AI data center buildout has boosted local economic activity and aggregate domestic demand, but its impact on GDP has been largely offset by a widening of the U.S. trade deficit (Chart 3).

261111ML2.jpg

261111ML3.jpg

Although the direct impact of the AI buildout on GDP growth has been muted, as Furman points out, the AI buildout has contributed to the ebullient stock market. The AI buildout has raised profit expectations and raised stock market expectations. This has fueled a positive wealth effect and boosted consumption growth. Accordingly, the total direct and indirect impacts of the AI buildout on economic activity has been larger than the measured impact of business investment net of the rise in the trade deficit.

Looking forward, the AI data center buildout will likely slow, but not come to a halt. As displayed in Figure 4, a sizable number of projects are under construction (particularly in Texas, Virginia, Pennsylvania, Ohio and Indiana), and these are unlikely to be abandoned. As shown for the different states, other AI projects have already been announced or are in the planning stage. It is uncertain whether they will move ahead; most likely, some will proceed with development while some will be cancelled. In the short run, this will subtract from economic activity. At the same time, however, it will reduce the demand for imported capital goods, which will provide an offset and significantly mitigate the net economic impact.

261111ML4 b.jpg

There are negatives and positives stemming from the shift in sentiment toward investment in AI data centers. The biggest negative is it would slow the implementation of productivity-enhancing AI. This will slow advances in economic performance, even if its impact on GDP is modest. It will reduce employment in construction and dampen economic activity in the regions where data centers would be built.

Three positives seem apparent. First, credit markets were beginning to have difficulty digesting the increasingly larger bond issuances by the big cap AI firms, with potential negative spillover effects on broader bond markets. A slower AI data center buildout would involve less bond issuances and fewer concerns about credit quality, potentially providing relief to markets. Second, the high global demand for capital goods required for the AI data center buildout has been absorbing capacity and exerting upward pressures on prices of select capital goods, and those pressures may ease.

Third and perhaps most importantly, history is marked by spurts in technological innovations that have involved excessive investment with pronounced losers along with profound technical advances. Recently, concerns about excesses in AI have mounted--in valuations, in debt, and elsewhere. The shift in political leadership that seems likely to slow the AI data center buildout may avoid some of the excesses. New innovations and investments in them always involve losers as well as winners. Let's hope that newly elected government officials and regulators act reasonably with a soft tap on the brakes that results in avoiding some of the excesses that have plagued historic investment booms while allowing the innovations to enhance productivity and standards of living.

The Economic Implications of the AI Data Center Buildout

Andrew Cates

The AI investment boom has become a source of demand for semiconductor exporters, equipment manufacturers and energy suppliers across the world. Its continued expansion increasingly depends on political consent as well as finance and technology. Resistance to data centres remains a modest immediate constraint, but delays could spread through supply chains and markets whose expectations assume rapid growth. For investors, the vulnerability lies in how widely exposure to the same spending cycle has spread.

The United States remains at the centre of that cycle. Its significance, however, is global. Asian producers supply chips and electronic components; European and Japanese manufacturers provide sophisticated equipment; and utilities and commodity producers are investing to meet the build-out's requirements. Pension funds, insurers and other investors share the financial exposure through global equity portfolios.

**A global investment cycle**
Four forces reinforce the expansion. Rising equity valuations improve financing conditions. Cheaper capital supports investment. That spending generates revenues and profits for suppliers, helping sustain the valuations that encourage further investment. The schematic traces this loop and the channels through which it reaches the wider economy.

261110AC1.jpg

Asia provides the clearest evidence of the international transmission. Semiconductors account for roughly two-fifths of South Korean exports, twice their share four years ago. Taiwan also has substantial exposure. Strong technology demand has supported more optimistic growth forecasts, but it has increased dependence on the investment decisions of a relatively small group of customers.

261110AC2.jpg

The supply chain extends well beyond chips. European and Japanese equipment makers benefit from semiconductor investment, while turbines, grid equipment, copper, steel and concrete are needed to build and power data centres. The benefits reach economies with very different industrial structures. Their exposure nevertheless has a common source: expectations that infrastructure spending will keep expanding.

That distinction matters. A slowdown need not involve a collapse in spending. Suppliers can face weaker orders and earnings growth simply because investment ceases to accelerate. Projects elsewhere in the supply chain may then become harder to justify.

**Profits and valuations reinforce the exposure**
US data illustrate the connection. Since late 2022, additional investment has been closely matched by rising profits in the sectors supplying equipment and services. This is an accounting relationship, not proof that every dollar of spending creates an equivalent profit. It nevertheless shows how closely the investment and earnings cycles have become linked.

261110AC3.jpg

High valuations support the process. Tobin's Q, which compares US equity values with the replacement cost of corporate assets, stands at roughly twice its long-run average. The measure requires caution: conventional replacement-cost estimates omit much intangible capital, and its relationship with investment has weakened as both have become more concentrated. Even so, a sustained market reversal could make further expansion harder to finance.

261110AC4.jpg

The financial consequences would be international. The weight of US equities in global portfolios exposes institutions and households far beyond America to the same earnings expectations. A correction could therefore weaken wealth and confidence alongside demand for the equipment and materials used in construction.

**Consent becomes an economic constraint**
Political resistance reaches the cycle through a different channel. Restrictions on data-centre construction can delay physical investment before their effects appear in market prices. US local opposition and campaigns in parts of Europe raise a broader question: can the industry secure consent quickly enough to meet its plans?

The immediate threat should not be overstated. Relatively few projects have been delayed, and investment commitments, jobs and financial concessions may yet secure approval. But longer planning timelines and more costly agreements could reduce the pace of expansion even without widespread prohibitions. Financing a project and obtaining permission to build it are separate requirements.

Strong profits offer some protection. Today's boom has a firmer earnings foundation than the late-1990s technology surge. Yet those earnings must remain sustainable, while computing equipment with an economic life of perhaps three or four years brings depreciation costs that will increasingly weigh on reported profits.

The central risk is a mismatch between the timing of AI's benefits and the expectations embedded in current investment plans. Slower spending, weaker supplier earnings and lower equity valuations could reinforce one another, transmitting the adjustment across borders. The scale is uncertain, but the breadth of exposure makes it a global macroeconomic issue.

Technological revolutions can deliver enormous benefits without rewarding every investment. For the world economy, the question is how smoothly the transition from infrastructure spending to broader productivity gains proceeds. Political consent now joins capital, computing power and electricity among the conditions shaping that transition. The benefits of AI may be widely shared; so, increasingly, are the risks of an interrupted build-out.

The AI Boom and the Price of Permission

Ethan Harris

One of my core views is that investors should consider the whole range of supply shocks, rather than focus on AI and the Iran war. The trade war is not over, immigrant reduction policy will continue to constrain labor force growth, and if a recent Fed study is correct, the El Nino weather event could boost US inflation by 1% next year.

**Team Transitory 2.0**

A popular view is that most of the recent inflation is transitory. Optimists argue that the pass through from tariffs to consumer prices is largely complete. Once the Iran war is over energy prices will tumble bringing down headline inflation and reducing the spillover effects on the core. On top of this, as Warsh and others argue, AI investments may be inflationary in the short run, but over time they will be a big boost to productivity, putting downward pressure on costs and inflation. If inflation will return to target on its own, the Fed does not need to be hiking rates.

261103EH1.jpg

**A full accounting**

I’m not drinking the Kool-Aid. I assume that the Iran war will eventually come to an end, and energy prices will come down, but I have no idea when that will happen. By my count, the Trump Administration has forecast a near-term end to the energy shock more than forty times since the war started. The timing also matters for AI. In the short run the AI buildout is inflationary and it could be a long time before, the disinflationary effects become clear.

The bigger story is that Team Transitory underestimates the persistence of other supply shocks. The supply of labor is very weak due to immigration reduction policy and the aging population. Both aren’t going away any time soon. There are a bunch of new tariffs in train, and an escalation of the trade war is likely after the election. The huge 6% of GDP budget deficit is also here to stay, drawing capital away from the private sector.

Finally, adverse weather events impact the productive capacity of the economy, both by disrupting activity and by diverting investment to mitigate their impact. My sense is that the inflationary impact of climate change has been small so far but will certainly build over time.

261103EH2.jpg

https://www.frbsf.org/research-and-insights/publications/economic-letter/2026/11/impact-of-el-nino-on-us-economy/

The immediate challenge is the El Nino weather event—a shift in trade winds in the Pacific that causes changes in weather. A new piece from the San Fran Fed looks at “[The Impact of El Niño on the U.S. Economy](https://www.frbsf.org/research-and-insights/publications/economic-letter/2026/11/impact-of-el-nino-on-us-economy/).” As the chart above suggests, this is the strongest event in many years. Using data for El Nino’s since the 1950s the study estimates that the current event added about 0.3% to the price level this summer and if its followed by a La Niña it will add about a percentage point to inflation next year (Chart below).

261103EH3.jpg

**The big picture**

With all the cross currents in the economy, it is important to look at the whole picture rather than focus on one or two stories. For example, Warsh’s “Task Force on Productivity and Jobs” only looks at the potentially most positive supply side shock—the implementation of AI. Warsh has said AI is “the most productivity-enhancing wave of our lifetimes—past, present or future.” Given the composition of the task force, I’m sure they will agree. However, the more relevant question is how AI fits in with the many other supply shocks hitting the economy.

Sorting through this complicated picture, my guess is that inflation will prove quite resilient. Warsh and the doves on the FOMC think the Fed only needs to remove a “dose” of accommodation to bring inflation down. They do not need to put “sand in the gears” of the economy. I disagree, bringing inflation down requires tight monetary policy to offset both strong demand and persistent supply shocks. Hiking the funds rate three more times, to 4.6%, could do the trick.

Even the weather is inflationary

Viewpoints: September 2020

Jump to: