Inauguruption: the flurry of Trump 2.0 executive orders

Before getting into the eruption of executive orders released yesterday, I want to address questions on a chart from our 2025 Outlook which received the most attention. Our goal was to highlight rising hyperscaler capital spending on AI infrastructure, and the general principle that these companies will need hundreds of billions in new incremental AI revenues to maintain their current margins due to these new outlays.

What was the inspiration for this chart?

“AI’s $600 billion question”, a piece by David Cahn at Sequoia Capital (previously at Coatue)

Are GPUs really 50% of total data center costs? I have read figures closer to 70% Estimates vary. I used a 50% GPU share as an estimate for a greenfield (new) data center based on comments from Jensen Huang in NVIDIA’s October 2023 investor presentation, and a chart from GenAI Experts, a group comprised of AI professionals from Meta, Uber, Scale AI, Goldman and PwC.

Since hyperscaler capital spending is capitalized, it wouldn’t immediately reduce their gross margins from an accounting perspective, right?

That is correct, the impact would show up over time via depreciation. However, these outlays do immediately affect free cash flow margins. Hyperscaler gross margins are roughly 2x their free cash flow margins, so had I used free cash flow margins as a benchmark instead, the annual revenue requirement would have been $350 billion per year. Still a lot more than the incremental new AI revenues that hyperscalers have been earning.

These outlays won’t continue forever, right? So why be so concerned about the need to earn a perpetual $400 billion in new revenues?

That’s a good question. As far as I can tell, most analysts covering the hyperscalers project multiple years of elevated capital spending on AI infrastructure. One analyst referred to such outlays as “table stakes”; in other words, the amount of spending needed to just remain competitive. In addition, projections from Lawrence Berkeley Labs show data center power demand doubling from its current 4.4% of US electricity generation, and Independent System Operators like PJM and MISO are scrambling to add new generation capacity based on what they’re seeing. So, it looks like we’ve got at least a few more years of elevated hyperscaler spending ahead.

Isn’t it too soon to worry about the need for “killer app” business solutions that will utilize all of this AI capacity? Isn’t that inevitable based on what we have seen so far from early stage GenAI applications?

The early-stage applications are certainly compelling. Last week, Goldman’s CEO said that 95% of an S-1 filing could be completed by AI applications in just a few minutes compared to 6 banker analysts spending two weeks drafting documents, and there are other examples like it. But this observation does not give us (yet) any visibility on to the price underwriters would pay for these applications, and how these amounts compare to AI capital outlays. That said, many analysts are not worried at all about corporate uptake of AI capacity. Examples include a response to the Sequoia piece in Forbes, and another from Foundation Capital.

The Inauguruption: the flurry of Trump 2.0 executive orders

Trump 2.0 is a hodgepodge of distinctly American political strains: the bare-knuckled nationalism and anti elitism of Andrew Jackson, the tariff-loving protectionism of William McKinley, the small-government/pro business policies of Calvin Coolidge, the unforgiving enemies lists and vendettas of Richard Nixon, the deportation policies of Dwight Eisenhower, the manifest destiny of James Polk and the isolationism of 1914-era Woodrow Wilson (yes, there are apparent contradictions in Trump’s agenda). American First policies create risks for investors since its supply side benefits collide with its inflationary tendencies; there’s not a lot of room for error at a time of elevated US equity multiples. It looks like it will be a volatile year based on changes so far in the 10-year Treasury, but there’s not enough negative information at this time to change strategy in portfolios positioned for continued US growth and outperformance, particularly given a more benign tariff rollout.

On the next few pages, we look at the eruption of executive orders yesterday with a focus on the ones with the largest impact on markets and the US economy.

Some notes on the Trump 2.0 hodgepodge:

• Trump’s manifest destiny. Trump mentioned expanding America’s footprint by retaking control of the Panama Canal, making Canada the 51st state, pursuing a purchase of Greenland from Denmark and planting an American flag on Mars. On the Panama canal: Panama was the first Latin American country to sign up for China’s Belt & Road Initiative. Chinese/HK firms now operate ports at both ends of the canal, a Chinese firm constructed a bridge across the canal and there’s discussion of adjacent Chinese rail and port projects. This casts some doubt on Panama’s ability to effectively safeguard canal neutrality as agreed in the treaty

• Nixon’s enemies list compiled by Presidential Counselor Charles Colson included a total of 220 politicians, celebrities, businessmen, reporters, labor leaders and academics. Many people on the list stated that they considered being hated by Nixon as their greatest accomplishment. You can view the original 1971 version of the Nixon enemies list here, in the University of North Carolina archives

• McKinley tariffs. Known as the Napoleon of tariffs, the Republican McKinley increased the average duty on imports to almost 50%, and to 70% on imported tin plates. The subsequent increases in inflation were politically disastrous: in the 1890 midterms, Republicans lost their majority in the House with seats falling from 171 to 88, and after the 1892 election, all branches of government were under Democratic control.

Immigration. Trump deportation policies might seem extreme1 and costly2, causing think tanks like the Peterson Institute to write about potential drags on growth and employment3. But like Newton’s Third Law of Motion, sometimes every political action has an equal and opposite reaction. The Biden administration oversaw the largest uncontrolled immigration surge on record, resulting in soaring unresolved immigration cases, an erosion in municipal solvency and scenes of lawlessness at the border. The Trump deportation policies are a reaction to that, and possibly temporary.

US labor markets may start tightening again: the JOLTS job opening rate is rising, the S&P Human Resources Index is rallying and small business optimism/capex plans are rising. Should labor supplies tighten excessively, the Administration would have the tools at its disposal to relax immigration restrictions. And remember: Trump stated again last night that he’s in favor of legal immigration, which could be increased as necessary.

The proposed end of birthright citizenship is the most controversial of Trump’s immigration-related executive orders. The order directs federal agencies to adopt an interpretation of the 14th Amendment's Citizenship Clause that conflicts with the past century of practice based on a 1898 Supreme Court case. It takes effect for children born at least 30 days from now. This will almost certainly be one of the first Executive Orders to make its way to the Supreme Court; it’s unlikely in my view that a Court majority would accept Trump's interpretation.

Energy and related infrastructure: read the fine print. You can try to streamline permits for pipelines and transmission projects with executive orders, but without explicit Federal eminent domain legislation, projects can still be challenged for a number of reasons. As shown below, oil pipeline, gas pipeline and transmission line projects in the US have practically ground to a halt. It would be quite a feat from a political perspective to jump start the pace. The Trump agenda also calls for an “end to leasing to wind farms”, but only 2% of US wind power takes place on public lands as opposed to private lands, so that’s a marginal issue at best.

Trump’s executive orders also paused distribution of undisbursed loan and grant monies from the infrastructure and energy bills (specifically mentioning EV charging stations as area for pausing disbursements) until Agencies come up with new energy projects consistent with Trump America First objectives. This would not count as impoundment as far as we can tell given the broad spending discretion defined in the original bills. These orders explain why in mid-January, Biden rushed through a $6.5 billion loan to Rivian and a $1.7 loan guarantee to Plug Power for a hydrogen project (insert face-palm emoji here).

Trump also refers to terminating the “electric vehicle mandate”. My interpretation: Trump is referring to Biden rules announced in March 2024 that reduce allowable fleet-level GHG emissions by cars and trucks by ~50% by 2032. Such rule changes could slow auto industry development of EV models, but the larger driver of EV adoption is probably the $7,500 EV subsidy included in Biden’s energy bill. As far as I can tell, Trump cannot unilaterally rescind the EV subsidy as it was part of prior legislation (the humorously named “Inflation Reduction Act”). EVs were less than 3% of the US passenger car fleet by the end of 2024, so I don’t see this as a huge economic driver just yet. The biggest possible casualty: all those EV battery assembly plants that are being built in red states from Georgia to Michigan.

Trump might have more luck achieving Bessent’s target for higher domestic energy production. Based on our energy math, Bessent’s 3-3-3 target amounts to a 7% increase in oil, gas and NGL production (most of which will continue to rely on hydraulic fracturing) and is a very achievable goal. Whether this brings down US energy prices is less clear. That depends on how much of the new production is exported, and the degree to which states continue to disconnect coal, gas and nuclear power in favor of renewables and energy storage. As shown below, inflation in consumer and industrial electricity supply chains is deeply entrenched at this point.

International taxation. I was glad to read this part of the America First manifesto if it means what I think it means: “America will no longer be beholden to foreign organizations for our national tax policy, which punishes American businesses”. In November 20235, I wrote about the Pillar II tax system that Europe was attempting to impose on the US. This was my conclusion at the time: “Pillar II provisions are essentially bounty-hunting rules that allow a third-party country to intervene in the tax policy of other countries and collect tax revenue that it has no legal nexus with. By allowing non-US countries to collect taxes from US companies on income earned in the US, Pillar II arguably undermines US tax policy and hamstrings Congress’ ability to design policy based on US needs”. The Biden Administration supported Pillar II and proposed changes to bring US corporate tax law closer to it, but tax policy can only be set via Treaty by the Senate or via Congressional legislation. So far, Pillar II rules have not been adopted by the US, and neither have Pillar I rules on Digital Service Taxes, another thinly veiled European attempt to impose penalties on US tech giants.

“Drain the Swamp/DOGE” executive orders include a hiring freeze, a pause in regulations not yet in effect, rescinding Biden-era executive orders, returning the federal workforce to the office, etc. As we explained in the 2025 Outlook, Federal employment of 3 mm people is at its lowest level as a share of US employment in 85 years (~2%). Within Federal workers, the largest employer is the Dep’t of Defense (excluding active military) followed by the Postal Service and Veterans Affairs. As for agencies in the DOGE crosshairs: the Environmental Protection Agency, Securities and Exchange Commission and Department of Labor when combined account for less than 1% of federal workers, while the Department of Education accounts for just 0.14%. So, it’s not clear that DOGE will be able to move the needle on government spending here.

Having lost his Sancho Panza (Vivek Ramaswamy is reportedly going to run for Governor of Ohio), DOGE Quixote (Elon Musk) will need to proceed on his own in his quest to reduce government spending. Repealing the Biden Executive Actions shown below would save around $100 billion per year, which is a small step towards the larger goals that DOGE has established for itself. Where might DOGE have to look after repealing Biden Executive Actions: entitlement spending and defense spending6, which are much larger drivers of sky-rocketing US debt levels than non-defense discretionary spending or the size of Federal government Agencies.

The biggest executive order surprise: the slow rollout of tariffs. Trump stated last night that 25% tariffs could be imposed on Mexico and Canada beginning on February 1 since they’re “allowing a vast number of people over the border”, but there were few mentions (yet) of explicit tariffs on China or a universal tariff. Trump also threatened 25% tariffs on Mexico and Canada in his first term but they were never implemented. Trump did sign an executive order directing federal agencies to examine unfair trade and currency practices and to assess whether foreign governments have complied with terms of existing trade deals.

I still expect Trump to increase tariffs on China and on EU auto imports as well. If there is a universal tariff, I expect it would only apply to critical imports which are 10%-20% of all US imports. As a reminder, most macro economists who study tariffs believe that they would reduce US manufacturing employment. If they’re right, the impact could well be felt in red states more than in blue ones (8 of the top ten import/GDP states are red).

While most economists believe tariffs are unlikely to restore US most manufacturing jobs lost to globalization (see chart, upper right), let’s go back to Newton’s Third Law of Motion and reactions/opposite reactions again. As shown below, the US allowed the international trade system to get to the point where most countries had higher tariffs on the US than the US had in exchange. By 2015, almost every country was below the “tariff reciprocity line”. Had the US not gotten to this point, and had US industrial production not stagnated since China’s entry into the WTO, we might not be talking about tariffs today. Like deportations, should tariffs prove to be the minefield that economists expect them to be, they can always be adjusted and/or repealed.

Lastly, on deregulation. Trump mentioned a requirement that 10 existing rules be eliminated for any new rule to be enacted. This follows on the 2-to-1 standard that Trump applied in his first term. But here comes the fine print again: it’s hard to dismantle the regulatory state. As explained below, the notice-and-comment rulemaking process could take 6 to 12 months, and sometimes the Courts side with those challenging new rules (or the rescinding of existing rules). During Trump’s first term, out of 77 major rules that were challenged, Trump won just 31% of the time, experienced a mixed outcome in 12% of cases and lost the rest of the time7. But even if the 10-to-1 standard doesn’t work as planned, it will still probably dampen the regulatory juggernaut, as similar policies did during Trump’s first term. One thing’s for sure: our CEO clients generally believe that the US economy would benefit from at least a temporary slowdown in the breakneck pace of regulation.

Source: Hong Leong Investment Bank Research - 22 Jan 2025