Liberation Day, One Year Later: The Tariff Reckoning

The One Thing: One year after Liberation Day, every metric the tariffs promised to move went the wrong direction — 89,000 manufacturing jobs lost, the trade deficit wider, households $1,700 poorer — and the Supreme Court already struck down the legal basis in February. The experiment is over. The question is what comes next.

If You Only Read One Thing: Tax Foundation's Liberation Day one-year review — comprehensive data rundown on what the tariffs actually delivered versus what was promised, without the political spin in either direction.

TL;DR: April 2 is the one-year anniversary of the largest tariff shock in U.S. history, and the data verdict is brutal. Also: Elon Musk filed confidentially for the largest IPO in history at $1.75 trillion — a Starlink+xAI combined stack that is the only serious non-Big-Tech AI distribution play. And the "Trump wins on rare earths" headline from the Beijing summit is wrong: China extracted a 7-month lease with an expiration date.

Liberation Day at One: The Tariff Experiment Has Results

Every major economic promise attached to the April 2, 2025 tariff announcement has failed to materialize. One year of data is now available, and the verdict is not ambiguous.

The stated goals were industrial revival, a shrinking trade deficit, and job creation. The actual results: 89,000 manufacturing jobs lost between April 2025 and February 2026, the goods trade deficit widened approximately 2% to $1.24 trillion, and the average American household paid an additional $1,700 in tariff-related costs. Federal Reserve Chair Jerome Powell attributed "elevated readings" in goods-sector inflation directly to tariff effects. The government collected $151 billion in tariff revenue in the first five months of fiscal year 2026 — nearly four times the prior year pace — but the National Taxpayers Union's one-year review documents that this was not profit; it was a tax on American importers.

Why it matters — Second-Order Effects framework: Tariffs are a price floor on imports, not a wall. When you raise import prices, one of three things happens: domestic producers expand to fill the gap, consumers absorb the higher cost, or demand contracts. The U.S. manufacturing base lacked the capital and labor to expand fast enough to fill the gap, so the second and third effects dominated. Imports actually rose 4% in volume (Americans imported more, at higher prices), while exports grew 6% — the opposite of the intended dynamic. The tariffs succeeded as a tax collection mechanism and as a political signal. They failed as an industrial policy.

The most consequential development is not the economic data but the legal one. In February 2026, the Supreme Court ruled 6-3 in Learning Resources, Inc. v. Trump that IEEPA — the International Emergency Economic Powers Act — contains no authorization for tariffs. Chief Justice Roberts invoked the Major Questions Doctrine: Congress cannot hand the executive "monumental" economic powers without explicit statutory language. The result: approximately $166 billion in tariff refunds now owed, and any future tariff regime requires an Act of Congress. This is not a minor procedural setback. It eliminates the unilateral tariff weapon as a negotiating tool unless Congress acts.

Room for disagreement: The strongest counterargument is that tariffs were always a means, not an end — leverage for negotiating real structural changes in U.S.-China trade. The bilateral rare-earths-for-semiconductor-access deal announced yesterday suggests the tariffs created conditions for a negotiation that wouldn't have happened otherwise. That's a plausible reading. But you'd need to show that the resulting deal was worth the $1,700 per household and the 89,000 manufacturing jobs — and the deal's terms, examined below, don't support that conclusion.

What to watch: Whether Congress passes explicit tariff authorization legislation before the SCOTUS mandate takes full effect. Roberts' ruling requires it; the political will to pass it is unclear. The PIIE analysis suggests a Congressional tariff bill would likely result in substantially lower rates than the Liberation Day regime, since legislators face the retail voters who paid the $1,700.

Musk Files for the Largest IPO in History

On April 1, SpaceX submitted a confidential IPO filing to the SEC targeting a $1.75 trillion valuation and a $75 billion raise — which would exceed Saudi Aramco's $29.4 billion in 2019 as the largest public offering ever. TechCrunch confirmed the filing, codenamed "Project Apex," with a 21-bank underwriting consortium. The formal prospectus is expected in April or early May, with the IPO targeting June 2026.

To understand why $1.75 trillion is even being discussed, you have to start with February 2026. CNBC confirmed the SpaceX-xAI all-stock merger at a combined $1.25 trillion valuation — the largest merger in history. SpaceX was valued at $1.0T, xAI at $250B, with the exchange ratio set at 0.1433 SpaceX shares per xAI share. The combined entity now holds Grok (the AI model), Starlink (the satellite internet network), and Falcon/Starship (the launch vehicles). The stated strategic rationale is "orbital data centers" — space-based AI compute facilities delivering 100 terawatts of capacity, targeting a June 2026 IPO specifically designed to fund this buildout.

Why it matters — Aggregation Theory: What Musk has assembled is structurally different from every other AI play. Google, Microsoft, Apple, and Amazon all operate AI models that depend on their existing platform relationships — app stores, search defaults, enterprise contracts, cloud customers. Their distribution advantages are real but bounded by antitrust exposure and regulatory scrutiny. SpaceX-xAI has a different shape: Starlink's satellite internet reaches any point on Earth independently of terrestrial ISPs, government censorship, or telco agreements. Grok sits on top of that distribution layer. If you believe the coming decade of AI value accrues to whoever controls distribution to the next billion users not yet served by Big Tech, then the Musk stack has a genuine structural claim. This is why $1.75T is not irrational, even for a company that has never reported net income after 23 years.

The numbers have to work, though. Starlink reported 10+ million subscribers and $10 billion in 2025 revenue, growing 40-50% annually. Projected 2026 revenue sits at $20-24 billion with EBITDA margins approaching 25%. At even a 15x forward revenue multiple — reasonable for high-growth infrastructure — Starlink alone justifies $300-360 billion. The path to $1.75T requires the orbital data center thesis to be real, which is a bet on technology not yet demonstrated at commercial scale.

Room for disagreement: The most pointed critique is PitchBook's Granda: the valuation is "justifiable" on paper but assumes continuous 40-50% Starlink growth, space-based compute materializing in a commercially relevant timeframe, and Starship commercialization opening new revenue streams before interest rates punish the growth premium. None of those is implausible; all three at once requires exceptional execution over a 5-year horizon. The 30% retail allocation — versus the typical 5-10% in mega-IPOs — suggests Musk is deliberately courting the retail-investor loyalty he's cultivated on X, which is rational financially but a warning sign for institutional price discovery.

What to watch: The formal prospectus, expected within weeks, will show whether SpaceX is actually profitable at the operating level (versus the net income line distorted by R&D) and whether xAI's Grok has any disclosed revenue. Those two numbers will tell us whether the $1.75T is a grounded sum-of-parts or pure narrative.

The Contrarian Take: China's Rare Earths "Deal" Is a Ransom Payment With an Expiration Date

The headline from this week's Trump-Xi Beijing summit is that Trump won a concession on rare earths: China will supply rare earth elements and permanent magnets to the U.S. as part of a new bilateral framework. The White House is framing this as a win.

Here's why that's wrong. China restricted rare earth exports starting in April 2025 — targeting precisely the elements (samarium, gadolinium, terbium, dysprosium, lutetium) that are irreplaceable in defense electronics, electric vehicles, and precision manufacturing. CSIS documented the immediate damage to U.S. defense supply chains. China waited twelve months while U.S. manufacturers scrambled, defense contractors filed hardship waivers, and the White House came to the negotiating table. Then, at the Xi-Trump summit, Beijing extracted two things in exchange for supply resumption: (1) the U.S. suspends its own export controls on rare earths until November 27, 2026, and (2) Washington signals willingness to loosen semiconductor export restrictions that had been targeting Nvidia and advanced AI chips.

According to China Briefing's summit analysis, China simultaneously ratcheted up export controls on April 2 while announcing the supply arrangement. That's not a coincidence — it's a negotiating posture that tells you the scarcity is still China's to manage. The supply arrangement runs until November 27. That's seven months. There is no structural solution in this deal: no new U.S. rare earth mining capacity, no allied sourcing diversification, no processing facility construction timeline. There is only a clock ticking toward the next identical crisis.

The semiconductor concession is the larger story. The U.S. export control regime — which had been preventing China from acquiring advanced AI training chips — was the single most consequential asymmetric advantage the U.S. held in the AI competition. Trading any portion of it for a seven-month rare earth lease is a poor exchange rate. CSIS' assessment notes the deal "restores access" but does nothing to address structural dependency. The word "restores" is doing a lot of work there.

What Bloomberg Missed

The SCOTUS tariff ruling and the rare-earths deal are the same story. Roberts' February ruling invalidated the IEEPA emergency power basis for tariffs, eliminating Trump's main unilateral trade weapon. The rare-earths deal's weak terms — a seven-month supply arrangement, no structural fix — are downstream of the fact that the U.S. lost leverage precisely as these negotiations were concluded. Bloomberg covered both stories separately; the connection is the actual insight.
The 30% retail allocation in SpaceX's IPO is structurally unusual. Major IPOs typically reserve 5-10% for retail. Musk is allocating 30% — three to six times the norm. This creates a retail demand floor that can suppress price discovery and compress institutional due diligence. It's a mechanism Musk used with X (Twitter acquisition, leveraged buyout with debt) and it reflects a deliberate choice to control the investor base composition, not just maximize valuation.

Quick Takes

Iran, Day 35: April 6 Is Four Days Away. IEA director Fatih Birol warned Tuesday that April oil supply losses would "double" compared to March, as stockpile drawdowns accelerate faster than rerouting can compensate. The UK is hosting a 35-country international conference this week on maritime security restoration — a meaningful escalation in allied coordination. Trump simultaneously claimed Iran's president requested a ceasefire; Tehran denied any such communication. The April 6 deadline for U.S. strikes on Iranian power infrastructure remains active. My April 1 prediction that the deadline would slip a third time still stands as the most likely outcome — but "most likely" and "certain" are different things when Trump is making decisions four days out. (CBS News)

Oracle Cuts 20-30K to Fund $156B AI Capex — The Starkest "Labor for Compute" Swap Yet. Oracle announced terminations for up to 30,000 employees (18% of global workforce), via 6 a.m. emails rolled out March 31 with no advance notice. The rationale: free $8-10B in cash to fund $156B in AI infrastructure commitments. The number that matters is not the layoff count but the backlog: Oracle has $523B in remaining performance obligations, up 433% year-over-year. This company is not shrinking — it is changing what work it needs done. The labor market implications for enterprise software developers are severe. (The Next Web)

Artemis II Launched Yesterday, First Crewed Lunar Mission in 54 Years. NASA's Artemis II lifted off April 1 at 6:35pm EDT on a 10-day free-return trajectory around the Moon — the first humans to travel beyond low Earth orbit since Apollo 17 in December 1972. Commander Reid Wiseman leads a crew including Victor Glover (first Black astronaut beyond LEO), Christina Koch (first woman to travel to the lunar vicinity), and CSA astronaut Jeremy Hansen (first non-U.S. citizen beyond LEO). The mission uses NASA's SLS rocket, not SpaceX — notable because SpaceX had its Lunar Gateway module contract cancelled in March 2026, one month before the mission it was supposed to support. (CNN)

Stories We're Watching

Iran, Day 35+: The April 6 Deadline (Day 5). The original framing of this conflict as a military closure is wrong — as I covered Tuesday, Iran is running a tiered-access blockade where Russian and Chinese ships transit freely. The April 6 deadline is for U.S. strikes on Iranian power infrastructure, not a reopening date for Hormuz. The question is whether Trump extends a third time (as I predicted) or whether he follows through on a strike that would immediately escalate oil prices. IEA's doubling-loss forecast for April is the new pressure variable pushing toward some resolution.

SpaceX IPO Prospectus: Watching for Net Income and Grok Revenue. The formal S-1 prospectus is expected within weeks. Two numbers will determine whether the $1.75T valuation holds: (1) SpaceX operating profit (versus net income, which is distorted by R&D capitalization), and (2) any disclosed Grok revenue from xAI's commercial API. If Grok is generating meaningful enterprise revenue — even $500M-$1B — the xAI merger starts looking smart. If Grok revenue is zero, the $250B xAI acquisition price was for a technology that hasn't yet found a product-market fit outside of X's captive user base.

The SCOTUS Tariff Ruling's Congressional Follow-On. Roberts' majority opinion in Learning Resources v. Trump requires explicit Congressional authorization for any future tariff regime. The administration has until the mandate takes effect to get Congress to act. Watch for a vote before the November midterms — any senator from a manufacturing state faces a difficult choice between protecting domestic industry and removing the tariff burden from their retail constituents.

The Thread

The stories today share a common theme: capital intensity at scale without proven ROI. SpaceX is asking markets to value a company that has never been profitable at $1.75 trillion on the bet that orbital data centers materialize. Oracle is firing 30,000 people to redirect cash into $156 billion of AI infrastructure whose customer base exists on paper in a $523B backlog but hasn't fully been built yet. The tariff regime collected $151 billion in one year — and the $166B in court-ordered refunds exceeds what was collected.

Every major capital bet this week is structurally sound if the underlying thesis is correct: AI infrastructure demand is real and growing, Starlink distribution reaches the underserved world, Chinese rare earth dependency is manageable in the short term. The risk is not that any individual bet is wrong. The risk is they're all being made simultaneously, by companies with overlapping balance sheets and overlapping customer bases, in a rate environment where the cost of being wrong is not a setback but a restructuring. Ben Thompson's Aggregation Theory explains who wins when AI demand materializes. It doesn't tell you what happens when everyone is building for the same winner at the same time.

Predictions

Prediction 1: The SCOTUS ruling in Learning Resources v. Trump forces Congress to pass explicit tariff authorization legislation before November midterms, resulting in a negotiated rate structure substantially below Liberation Day levels — likely a 10% universal baseline with targeted sector-specific carve-outs. Congress cannot politically defend the current IEEPA-void state heading into a midterm where 71% of Americans want more trade oversight (Pew, Q1 2026). Confidence: medium. Check date: June 30, 2026.

Prediction 2: SpaceX will price its June IPO below $1.75 trillion — in the $1.2-1.5T range — as institutional investors force a haircut on the orbital data center thesis. The historical precedent: Saudi Aramco filed at $2T and priced at $1.7T after institutional pressure. SpaceX is less mature, carries zero net income, and the xAI merger premium is unearned until Grok shows commercial revenue. Retail enthusiasm (driven by the 30% allocation and Musk's X platform) will not be sufficient to hold the full Ask without institutional support. Confidence: medium. Check date: June 30, 2026.

Generated: 2026-04-02 | Daily News Briefing | ali.mujtaba.lakdawala@gmail.com