Energy Musings - March 23, 2026
Data center construction and power needs have shown a huge need for more construction and manufacturing workers. The maritime industry will need more workers, too. Are there enough?
Labor: Overlooked Hurdle For Maritime Revitalization
A column by financial consultant John Mauldin caught our attention. It included a discussion of the construction boom for data centers and Artificial Intelligence (AI). This has been a hot topic because of the impact of the surge in new data centers on the nation’s power grid and consumer electricity bills.
In 2025, U.S. Gross Domestic Product, the sum of all our economic activity, grew by 2%. According to investment advisor Peter Boockvar, over half that growth was from data center construction. The trend will be higher in 2026 and likely in 2027 as well. How do we know? Wall Street tells us.
The four largest technology companies – Amazon, Google, Meta, and Microsoft – announced plans to invest $700 billion in AI infrastructure. To support their spending plans, some of the largest bond offerings in recent years have been floated. These spending plans are part of the largest technology spending in history, much more than was spent in the dotcom era at the end of the 1990s.
In an article last summer, consultant McKinsey estimated that by 2030, technology companies will invest almost $7 trillion in capital expenditures on data center Infrastructure globally. More than $4 trillion will be invested in computing hardware, with the balance allocated to areas such as real estate and power infrastructure. More than 40% of this spending will be invested in the United States.
Data center infrastructure will consume $7 trillion in capital expenditures through 2030.
McKinsey modeled three scenarios for the growth in data center power needs. Between 2023 and 2030, the consultants see a potential compound annual growth rate of 18% to 27% in gigawatts. It is this potential growth rate in power that has utilities and regional power grids fearful they will be stretched beyond capacity, leading to power rationing or blackouts. Grid expansions and additional transmission capacity will be needed.
Global demand for data center capacity could more than triple by 2030.
Adding to the challenge for power companies and grids is the steady expansion across the country. Data center developers are widely distributing new centers in regions with abundant power and less-strained grids. However, as more of these data centers begin operating, that status could change.
Data centers are being located in remote areas where power is abundant, and grids are less strained, for now.
Another trend in data center construction is the shift from training centers to inference centers. According to Tech-Insider.org:
“The cost of training frontier models, while still enormous in absolute terms, has become a smaller fraction of total AI compute spending. The real expense — and the real business opportunity — lies in serving those models to billions of users. Inference now accounts for an estimated 60 to 70 percent of total AI compute demand across major hyperscalers, up from roughly 40 percent in 2024.
“This shift matters for several reasons. Inference workloads have different hardware requirements than training. While training benefits from massive parallel processing across thousands of interconnected GPUs, inference prioritizes low latency, high throughput, and energy efficiency across distributed data centers. This creates openings for specialized inference chips — like Amazon’s Trainium and Inferentia, Google’s TPUs, and dedicated inference accelerators from startups like Groq and Cerebras — to compete with NVIDIA’s dominance in ways that were not feasible in the training-dominated era.”
This shift in data center construction is leading to an increase in manufacturing jobs. The upturn is seen in the latest data from the JOLTS Job Openings index.
The shift in data center construction means more manufacturing jobs.
Besides the increase in manufacturing jobs from AI, the upturn in job openings reflects the Trump administration’s efforts to reshore supply chains that were outsourced during the era of globalization. Another breakdown of construction jobs shows that while construction jobs are in decline, those tied to data centers have more than offset the decline.
More data center construction jobs are supporting the economy.
In a study by investment bank Goldman Sachs, the bank estimated that by 2032, the nation will need 300,000 new jobs to power new data centers. Additionally, another 207,000 jobs will be needed to build out the nation’s transmission system to deliver power to the new data centers. That is 500,000 new jobs in the utility sector alone.
A significant number of new workers will be needed to build out the power grid.
Mauldin’s column sought to answer the question of whether the economy was half full or half empty. In one case, you would be pessimistic about our economic future, while the other conclusion would make you optimistic. AI spending may help answer the question.
Mauldin quoted engineer, physician, and entrepreneur Peter Diamandis, who noted, “[T]his is beginning to generate serious revenues. $50 billion data centers generating $10 billion annually. $1 trillion in infrastructure deployment. Neuromorphic chips making AI 1,000x more efficient. Fusion energy and small nuclear reactors moving from experiment to grid power. AI is accelerating longevity research. Humanoids going from demos to war zones. SpaceX making orbital access routine.”
We agree with Mauldin’s observation that “We have not experienced anything like this that I can tell in human history.” Major technological revolutions, such as the steam engine, electricity, and automobiles, significantly impacted the economy. However, they did it over decades. Instead, this technology revolution is impacting in years. It will create challenges. There are negatives, such as jobs under pressure, while others are in demand. Mauldin’s concern is that the shift is happening so quickly, which he believes is a problem. Why, because decisions must be made with imperfect knowledge.
Last year, the U.S. trade group Associated Builders and Contractors (ABC) predicted that 439,000 additional workers would be required to meet demand for construction services this year and 499,000 in 2026. The ABC model uses historical data from the Census Bureau and the Bureau of Labor Statistics, which assumes that 3,550 jobs are created for every billion dollars spent on construction.
More construction spending means more jobs created.
Anirban Basu, ABC’s chief economist, said: “We still have a long way to go to shore up the talent pipeline. The data on the number of young people choosing a career in construction suggests that employing practical technology and innovation in educational programs and on jobsites helps maximize the productivity and efficiency of the construction workforce.
“There are also factors that could render this model overly conservative, meaning worker shortages could be more severe than predicted in 2025.
“Although the consensus forecast has construction spending increasing by less than 3% in 2025, that same forecast has underestimated growth by a significant margin during each of the past three years. If inflation dissipates in [the] coming months, borrowing costs will subside, and construction volumes will increase.
“Faster-than-expected immigration over the past few years has also bolstered labor supply, and potential changes to immigration policy will likely constrain worker availability.”
A critical issue for the construction industry is the high rate of retirement among skilled construction workers. The average age of a construction worker in the US is now over 42, and as seasoned professionals leave the industry, fewer replacements are entering the field at a sustainable rate.
Younger workers are gravitating toward white-collar and tech-based careers. Additionally, a strong societal emphasis on four-year college degrees over vocational training has reduced the pipeline of skilled tradespeople. These trends became more evident following the pandemic, as the accelerated labor force exits coincided with many workers seeking new careers or retiring.
Supply chain disruptions and project delays led to inconsistent work opportunities, discouraging new workers from pursuing careers in construction. Additionally, construction jobs are associated with unstable employment, challenging working conditions, and limited career mobility, making them less attractive.
Why is this discussion of construction employment important for the shipping industry? The key pillars of the Maritime Action Plan of the Trump administration are to rebuild the U.S. shipbuilding capacity, reform workforce education and training, protect the existing maritime industrial base, and support national security and industrial resilience. The pillar on workforce education and training focuses on mariners, a population that needs to grow, especially as the U.S. expands its shipping fleet. However, expanding the shipbuilding capacity means finding and training thousands of additional skilled construction workers, those who are currently in demand for building data centers and new manufacturing plants.
Shipbuilding employment will need to increase significantly.
While the shipyard employment data only extends through 2024, it is clear that U.S. shipyards face a worker shortage. There were 152,000 workers in 2024, 30,000 fewer than in 1988. Shipyards have been underinvesting in worker productivity, especially in those shipyards dedicated to building large, ocean-going vessels. South Korea’s Hanwha, which owns Philly Shipyard and has received orders for new ships, plans to send local workers to its Korean yards to train them in modern shipbuilding practices and technologies before sending them back to Philadelphia to begin constructing the new ships. Hanwha has also pledged significant investment to improve manufacturing efficiency.
Shipyard production has risen steadily since 1995.
What we cannot discern is how much the current ship and boatbuilding industry’s industrial production is affected by the smaller shipbuilding yards, which we know are very active. Again, we cannot tell how much the smaller yards have impacted the new orders data, which has shown a slowly rising trend throughout its history.
Shipyard orders have risen very slowly over time.
The current SHIPS Act still lacks champions within the administration and Congress, so its progress through the legislative process has been very slow. The Trump administration has promised legislative proposals in support of its Maritime Action Plan when it submits its FY2026 budget. The oxygen in Congress has been consumed by the partial shutdown over funding the Department of Homeland Security, the debate over the SAVE Act, and the Iranian conflict. Hopefully, there will be sufficient oxygen when the legislative proposals are presented.
Individual shipyards can do their part in boosting the skilled workforce, but companies will be reluctant to invest significantly if they believe they might lose the new workers to competitors. We have multiple sectors clamoring for more skilled workers, without which their growth plans can be derailed. We will be looking closely for plans to expand the shipyard’s skilled workforce, because without an overall plan, efforts to revitalize our maritime industry may be sidetracked.