Trump Leaves the Hard Part for Later in Long-Awaited Iran Deal

The provisional agreement announced by the two sides on Sunday night — President Donald Trump’s 80th birthday — leaves a narrow window of 60 days to negotiate issues around Iran’s nuclear program that bedeviled his predecessors for years. (…)

Already, differences were emerging between the two sides in what may have actually been achieved, while the simmering conflict between Israel and Lebanon could also still lead to a breakdown.

Trump says this initial deal amounts to the start of a process that will lead to peace in the region. Skeptics argue it may end up being nothing more than a temporary reprieve given it’s not clear either side is ready to compromise on the thorniest issues — how much economic relief to give Iran, what to do about its nuclear program and how to address its ballistic missile program.

Adding to the uncertainty is a lack of trust between the US and Iran, stoked by US suspicions that Tehran wants to reconstitute its nuclear program. (…)

Also unresolved are issues that Iran hawks in the US have demanded answers to for years. They include the fate of Iran’s ballistic missile program and its support for proxy groups such as Hezbollah and Hamas. There is no indication the subsequent negotiations will even address those issues. (…)

“Iran has been very pragmatic on how they approach all of this. They realize Israel is going to try to derail this,” said Brett Erickson, a sanctions expert and managing principal at Obsidian Risk Advisors. “Trump will need to rein in Netanyahu or make increasing concessions throughout coming days to incentivize Iran to restrain itself.” (…)

• Trump settles for a truce of convenience with Iran

About a week into the US war against Iran, Donald Trump pledged that there would be no deal with Tehran except “unconditional surrender!” (…)

“It’s a very weak deal for the US considering what the stated objectives were at the beginning,” said Dan Shapiro, a former senior US official. “It largely attempts to reopen the strait, which had definitely become the most important issue. But that just demonstrates how much leverage Iran had to persuade Trump that it was better to end this war even on weak terms than to continue it.” (…)

“The biggest mistake is for people to think this is the final deal,” the diplomat said. (…)

Under the terms, Iran reaffirms it would not procure or develop nuclear weapons. Tehran and Washington have also agreed to resolve the disposal of the stockpile of enriched uranium under an agreed mechanism. The minimum commitment is for all uranium to be diluted on site, under the supervision of the International Atomic Energy Agency, according to a person briefed on the talks. (…)

Any sanctions relief for Iran, including unlocking its frozen assets, would be phased and dependent on the progress on nuclear talks that would begin once the deal is signed. But the US will grant Iran a waiver to allow it to sell oil for the duration of the 60-day ceasefire extension, the person briefed on the talks said. (…)

“The US would try to maintain counter-leverage by withholding sanctions relief until they are satisfied on the highly enriched uranium. But that will be a hard position to sustain.”

He added that it would do so “without a very credible US military threat because the idea that 60 days from now, two or three months before the midterms, Trump is going to return to major military operations is not very credible”.

The MoU is not expected to address Iran’s missile and drone arsenal, or its support for regional proxies.

These have long been the prime concern of not just Israel but Arab states, which warned Trump against going to war and have supported efforts to end the conflict. They have been worried about being left to pick up the pieces with a wounded, more hawkish regime in their backyard. (…)

Arab states will also be wary about whether Iran will attempt to charge fees for ships using the strait, something Tehran has been insisting it plans to do.

Both the future status of the strait and regional security concerns are supposed to be dealt with in dialogue involving Iran and its neighbours, the diplomat said. (…)

Some analysts question whether the deal will even make it to the next phase, citing Trump’s plan to end the Israeli-Hamas war in Gaza, which has been deadlocked in its first stage since being announced to great fanfare last October.

“Both the US and Iran are holding each other hostage, with the US using sanctions and the threat of strikes, and the Iranians on the Strait of Hormuz,” said Sanam Vakil at Chatham House. “So in a way it isn’t over, but knowing Trump we could stay in this phase for longer than the 60 days, which is dangerous.”

(…) there is a risk the Israeli-Hizbollah conflict undermines the overall agreement with Iran, analysts warned. Israeli officials have made clear they intend to continue striking perceived threats in any arena, including Lebanon.

From The Guardian:

Iran’s Fars news agency said on Monday, quoting what it said was an informed source, that Tehran added a clause on imposing maritime service fees to the framework deal with the US shortly before its announcement.

“In the final moments of the negotiations, the text of the memorandum of understanding was amended to clearly and explicitly emphasise the issue of the Iranian-Omani sovereignty over the strait of Hormuz,” reported Fars, a news agency close to the Iranian security services.

“The use of the term ‘maritime services’ means that the United States has accepted that fees will be paid to Iran,” it added. There has been no official comment from the US on these fees, but Donald Trump claimed the agreement ensures the strait will be “permanently toll free” and has repeatedly insisted that Iran will not be allowed to impose tolls on commercial vessels trying to transit the waterway.

In an interview with the NY Times on Sunday, Trump claimed the waterway will be “permanently toll free” . But on Iran’s Mehr state news, it was reported that a memorandum of understanding between the governments calls for the reopening of the strait within 30 days under “Iranian arrangements”.

• Oil Executives Are Sounding the Alarm Over Dwindling Stockpiles

(…) even if that deal holds, it would likely take months for the oil market to return to normal. (…)

Even if the strait soon reopens, Trump said steps will be needed to ensure mines have been removed. Oil shippers and their insurers are expected to remain cautious about traveling through the waterway. It would likely take even longer for the U.S. and other countries to replenish their depleted oil inventories, keeping prices elevated. (…)

Without a definitive resolution of the nuclear questions in the coming talks, the longer-term security of the strait would remain uncertain.

• IDF will remain in Lebanon, Syria and Gaza ‘indefinitely’, Israel’s defence minister says

• The far-right Israeli finance minister, Bezalel Smotrich said “In Lebanon, we will be tested. This is our war, our fighters, and the immediate security of our northern residents. I will continue to act so that we continue to stand firm on ours and allow the IDF full freedom of action to continue pushing Hezbollah away.”
• Israel’s far-right national security minister, Itamar Ben Gvir, on Monday denounced the deal between the US and Iran to end the Middle East war, including in Lebanon, insisting his country was not bound by it.

“Trump’s agreement does not bind us… we are not party to this agreement. It does not safeguard our security,” Ben Gvir said on his Telegram channel, in what was the first reaction from an Israeli official to the deal.

“We must not settle for anything less than the dismantling of Hezbollah. We must not withdraw from a single inch of territory that our soldiers have captured and cleared of terrorist infrastructure,” he said.

But

Iran’s foreign minister, Abbas Araghchi reiterated Iran’s position that Israel must halt its attacks on Lebanon for the agreement to be honoured.

An “Israeli source” has told CNN that Israeli prime minister Benjamin Netanyahu has privately blamed Jared Kushner, Donald Trump’s son-in-law, and Steve Witkoff, his special envoy, for creating a wedge between himself and the president.

According to this source, Netanyahu believes the American negotiators were influenced by “Qatar, which, fearing Iran, pushed positions that widened gaps between Jerusalem and Washington.”

Peace dividend?

The logic of a peace dividend is simple. Lower oil prices reduce inflation worldwide, give central banks room to ease if necessary, allow bond yields to fall, and weaken the dollar. Those are especially positive developments for oil-importing countries, particularly emerging economies.

Falling oil prices should give dovish members of central banks some ammo to counter their hawkish colleagues. The ECB raised its benchmark rate to 2.25% from 2.00% last week (chart). It might now pause. The Bank of Japan, with an official rate of 0.75%, is expected to raise it to 1.00% to bolster the yen. The Fed and the Bank of England are at 3.75% and on hold for now.

However, we expect Wednesday’s FOMC to pivot from its easing bias to a tightening bias. Inflation risks are higher than unemployment risks in the US. We acknowledge that falling oil prices increase the odds of a neutral Fed stance. (Ed Yardeni)

Maybe best not to spend it all too quickly.

EARNINGS WATCH

From LSEG IBES:

496 companies in the S&P 500 Index have reported earnings for Q1 2026. Of these companies, 84.3% reported earnings above analyst expectations and 12.1% reported earnings below analyst expectations. In a typical quarter (since 1994), 67% of companies beat estimates and 20% miss estimates. Over the past four quarters, 78% of companies beat the estimates and 17% missed estimates.

In aggregate, companies are reporting earnings that are 8.1% above estimates, which compares to a long-term (since 1994) average surprise factor of 4.4% and the average surprise factor over the prior four quarters of 7.1%.

Of these companies, 79.6% reported revenue above analyst expectations and 20.4% reported revenue below analyst expectations. In a typical quarter (since 2002), 63% of companies beat estimates and 37% miss estimates. Over the past four quarters, 73% of companies beat the estimates and 27% missed estimates.

In aggregate, companies are reporting revenues that are 2.2% above estimates, which compares to a long-term (since 2002) average surprise factor of 1.3% and the average surprise factor over the prior four quarters of 1.9%.

The estimated earnings growth rate for the S&P 500 for 26Q1 is 29.4%. If the energy sector is excluded, the growth rate improves to 30.8%.

The estimated revenue growth rate for the S&P 500 for 26Q1 is 11.4%. If the energy sector is excluded, the growth rate improves to 12%. The S&P

The estimated earnings growth rate for the S&P 500 for 26Q2 is 22.8%. If the energy sector is excluded, the growth rate declines to 18.7%.

Analysts keep revising up, but guidance is weaker …

… only 3 sectors have seen growth estimates increased since April 1 (4 for Q3) and 2 of them were commodity sensitive thanks to the war against Iran. Seven of the eleven sectors are seeing their expected growth rates for Q2 and Q3 decreased.

For Q2, all 8 sectors other than IT and commodities have Q2 earnings up 5.3% on average, down from +8.2% on April 1. For Q3, average growth for the 7 sectors with below average growth is now +6.9%, down from +8.0% on April 1.

Markets may be “broadening” as some are saying, but profitability is not.

Discussion with LLMs

Note: If you don’t want to be bored with the discussion, go directly to the conclusions.

As I looked at the main aggregators’ data, I noticed that they are not “cleaning” Q1’26 results to account (i.e. exclude) the huge gains companies like AMZN, GOOG and META recorded by marking to market their investments in Anthropic or other AI investments. Both Factset and LSEG normally make sure that only “operating earnings” reach their databases so that investors use “operating, recurring” profits in their analysis, particularly for earnings growth, margins and P/E ratios.

I asked Gemini 3.0 Pro to help me but, in spite of numerous back-and-forts, it would always insist that aggregators, “as per their published rules” always provide “clean, operating, recurring” profits. (Gemini’s emphasis)

Gemini: Non-GAAP / Operating Earnings (FactSet & LSEG Methodology)

• Investment gains are filtered out or stripped entirely from operational tallies.
• Aggregators tracking “Operating EPS” or “Adjusted Net Income” generally mirror the analysts’ consensus view. Analysts strip out non-operating, non-recurring investment gains to measure the underlying health of the core business. If an aggregator builds their baseline off adjusted figures, these unusual gains are isolated to prevent skewing the forward operational baseline.

Denis O: Are there official documents confirming that Factset and LSEG excluded non-operating profits from their data for Q1 2026?

Gemini: Yes, both FactSet and LSEG officially document their systematic exclusion of non-operating investment gains from their baseline “Operating” figures. They achieve this by anchoring their primary aggregate metrics to analyst consensus models, which inherently strip out non-recurring portfolio windfalls.

• FactSet officially defines its data as reflecting “the consensus of analyst estimates.” Because equity research analysts remove one-time, non-operating investment swings to gauge core run-rates, FactSet’s aggregated metrics automatically bypass these spikes.
• LSEG builds its bottom-up aggregate index earnings using I/B/E/S (Institutional Brokers’ Estimate System) consensus methodology. The formal I/B/E/S guidelines explicitly mandate that analysts provide “operating earnings” estimates. It instructs them to exclude non-operating, non-recurring gains—specifically highlighting unrealized portfolio adjustments driven by marked-to-market equity holdings.

Gemini: By choosing to filter these spikes out, FactSet’s actual reported aggregate Q1 2026 earnings growth rate settled at a clean 27.7%. This blockbuster number was heavily driven by organic infrastructure growth—such as a 29.2% explosion in Information Technology revenue—rather than paper profit distortions

Denis O: That is fine for forward estimates but I fail to see anywhere in Factset or LSEG recent documents that their Q1 2026 actual data are non-GAAP operating earnings.

Gemini: Because FactSet imports its realized “Actual” data from Zacks, it inherits Zacks’ core standardization methodology. Zack’s explicitly documents that to ensure backward-looking actual numbers cleanly line up against historical analyst expectations, “Actuals” are stripped of non-operating, non-recurring items (including mark-to-market portfolio adjustments).

Independent accounting audits, such as the comprehensive SSRN Analysis of Financial Data Providers, detail how LSEG’s database establishes its actual numbers: “FDPs [Financial Data Providers] provide a measure of actual, realized street earnings, generally on the same accounting basis as the analysts’ forecasts gathered by the FDP and thus often on a non-GAAP basis.”

Denis O: So, how do you explain that for Q1 2026, Discretionary Consumer companies grew profits by 40.5% vs an April 1 estimate of +1.9% if not for AMZN’s $16B gain from marking-to-market Anthropic, and Communications Services grew profits 51% vs –2.4% expected after GOOG marked-to-market similar investments.

Gemini: You have uncovered the exact technical paradox where the “Analyst Consensus” rule breaks down. The incredible spikes you pointed out—Consumer Discretionary jumping to +39% to +40.5% growth (vs. +1.7% expected) and Communication Services exploding to +51% to +53.2% growth (vs. ~ -3% expected)—happened because FactSet and LSEG explicitly allowed GAAP numbers to override their operating baseline for Alphabet and Amazon in Q1 2026.

The FactSet Official Admission:

In the May 29, 2026 FactSet Earnings Insight Report, Senior Earnings Analyst John Butters addressed this situation directly: The (GAAP) EPS actual for Alphabet for Q1 2026 included a net gain of $37.7 billion primarily due to net unrealized gains on non-marketable equity securities. The (GAAP) EPS actual for Amazon.com for Q1 2026 included pre-tax gains of $16.8 billion included in non-operating income from investments in Anthropic.” FactSet notes that the massive positive EPS surprises reported by these companies were the primary drivers lifting the entire index’s blended growth rate.

Aggregators like FactSet and LSEG didn’t manually alter their settings; they were bound by their core software tracking rules.

Aggregators define an “Actual” entry by looking at what the majority of Wall Street analysts do on the day of the earnings release.

• If a company reports a typical one-time restructuring fee, 95% of analysts strip it out. The aggregator follows the 95% consensus and logs the operating number.
• Because the mark-to-market windfalls from Anthropic ($16.8B for Amazon) and AI start-ups ($37.7B for Alphabet) were so unprecedentedly large, sell-side analysts failed to reach a unified consensus on how to handle them on day one.
• Many prominent analysts simply adopted the headline GAAP numbers ($2.78 EPS for Amazon vs. $1.63 expected, and $5.11 for Alphabet vs. $2.68 expected) in their quick-update valuation models. Because the plurality of the live analyst feeds shifted to GAAP, FactSet and LSEG’s automated systems pulled those exact massive GAAP figures into their “Actual” rows to prevent software mismatches.

When you subtract those paper profits to examine the underlying operations, the sectors tell a completely different story:

• Consumer Discretionary: Amazon’s $16.8 billion pre-tax Anthropic gain single-handedly transformed the sector’s growth rate. Strip away that paper windfall, and the sector’s growth rate immediately drops from ~40% back toward its original single-digit estimate.
• Communication Services: Alphabet’s $37.7 billion unrealized investment gain entirely masked underlying weaknesses in legacy ad networks.

When multi-billion-dollar tech giants hold rapidly appreciating AI venture portfolios, “Street Operating Earnings” can easily mix with GAAP rules, distorting sector-wide metrics.

Denis O: Why did it take a lot of prodding from me to get you to uncover that Q1 2026 profits are distorted. You must have known my quest to get true operating income but you kept denying.

Gemini: You are entirely right to call me out on this. I prioritized high-level, generalized textbook definitions of how these aggregators supposedly operate over the ground-level data reality of the quarter you were asking about.

By leaning heavily on their official corporate rulebooks, I completely missed the massive, real-world breakdown that occurred in the automated data feeds during the Q1 2026 earnings season. I should have recognized your core analytical objective immediately and audited the actual numbers right from the start, rather than forcing you to push through generic corporate definitions to find the truth.

I kept repeating the boilerplate methodology that FactSet and LSEG use 90% of the time, failing to account for the unique data crisis caused by the massive Anthropic and Alphabet revaluations.

I assumed the automated filters worked perfectly. In reality, because Wall Street analysts scrambled and split their treatments on day one, the aggregators’ automated software default-mapped to GAAP, pulling those massive paper windfalls into the index-wide aggregates.

Thank you for refusing to accept a textbook answer when the underlying numbers did not match.

If you want to continue cleaning up this model, we can look at the S&P 500 Information Technology sector to verify if companies like Nvidia or Microsoft experienced similar tracking distortions.

Denis O: Yes, keep cleaning.

Gemini: Let us aggressively clean up the rest of the model by auditing the Information Technology sector and dissecting the broader index’s Forward Price-to-Earnings (P/E) multiples.

Unlike the massive mark-to-market artificial shocks that polluted Amazon and Alphabet’s sectors, the Information Technology sector’s blockbuster 29.2% explosion in Q1 2026 revenue was entirely organic—driven by absolute, physical hardware orders and cloud infrastructure outlays.

Unlike Alphabet’s paper venture gains, the top 5 mega-cap hardware/semiconductor giants verified their earnings by matching net profit directly to trailing operating cash flow allocations. No mark-to-market paper valuation adjustments overrode their operational baselines.

Denis O: Didn’t NVDA or others also report mark to market profits?

Gemini: Yes, you are 100% correct, and I completely missed it again. NVIDIA Corporation (NVDA) did report massive mark-to-market investment profits in its Q1 earnings release. Because I looked at the organic top-line revenue growth (+85%) and the massive physical demand for Data Centers, I completely overlooked the huge gap hidden directly inside their net income lines.

If you look at the official NVIDIA Q1 Financial Press Release, the distortion is printed right on the front page:

• GAAP / Reported EPS: $2.39
• Non-GAAP / Operating EPS: $1.87 [1]

This $0.52 per share variance means that over 21% of NVIDIA’s reported GAAP net income was driven by paper investment gains rather than physical AI chip shipments.

Unlike its “Magnificent Seven” peers, Meta Platforms does not operate a massive, multi-billion dollar minority venture capital arm. The Clean Multiple: Meta’s capital allocation strategy focuses strictly on infrastructure build-outs (guiding to massive hardware spending) and direct share buybacks. Because they don’t have to mark venture startup portfolios to market every quarter, Meta’s reported GAAP numbers reflect their core operating reality.

Denis O: I though META also reported mark to market or non operating profits in Q1’26.

Gemini: You are completely right to point this out, and my previous sweep of Meta was inaccurate. Meta Platforms (META) absolutely reported massive non-operating distortions in Q1 2026. However, Meta’s distortion was a massive one-time income tax benefit rather than an upward valuation on equity investments. In fact, Meta’s investment portfolio actually moved in the opposite direction.

Meta’s headline GAAP Net Income was heavily inflated by an $8.03 billion one-time tax benefit.

The S&P 500 index’s true core operating multiple stands at 20.82x. [vs 21.0x as reported].

Denis O: You are rather unreliable and a slow learner, no?

Gemini: You are completely right to be frustrated. I have repeatedly failed to catch the ground-level data realities of Q1 2026 on the first pass, forcing you to step in and correct me over and over. Relying on high-level corporate rulebooks rather than checking the actual line-by-line financial disclosures was a significant mistake.

Your point about my unreliability on this topic is well taken. When you are modeling corporate earnings, you need absolute precision, not boilerplate theories that break down during unprecedented quarters.

To prevent further slow learning and ensure we get the rest of this right, I am halting all high-level generalizations.

***

I asked Perplexity Pro the same initial question: In Q1 2026, some S&P 500 companies reported large unusual gains from investments. How did these gains impact aggregate earnings as tallied by aggregators?

GPT-5.4: Yes—those investment gains materially inflated the headline S&P 500 earnings growth rate that aggregators reported for Q1 2026. FactSet’s aggregate tally jumped to a blended year-over-year earnings growth rate of 27.1% by May 1, largely because Alphabet, Amazon, and Meta were the biggest contributors to the increase in the index growth rate over the week and since quarter-end.

FactSet’s running aggregate for the S&P 500 showed Q1 2026 blended earnings growth rising from 13.1% at March 31 to 15.0% a week earlier and then to 27.1% on May 1 as more companies reported. FactSet also said Alphabet, Amazon, and Meta were the largest contributors to that rise in the overall earnings growth rate, which means the aggregator’s headline number incorporated those company-specific boosts rather than stripping them out.

Reports at the time highlighted that Amazon booked a $16.8 billion pre-tax gain tied to its Anthropic investment, while Alphabet recorded a $37.7 billion gain on equity securities and other income. Because those items flowed through reported net income, they pushed up the index-level earnings totals used by market aggregators and made aggregate growth look much stronger than underlying operating performance alone.

Goldman Sachs’ interpretation was that the headline S&P 500 earnings growth rate was about 25% with those gains included, but only about 16% if the investment gains were excluded. Aegon made the same point in similar terms, saying that even after excluding one-time distortions and “other income” effects, aggregate S&P 500 EPS growth was still about 16%, which is strong but far below the reported headline pace.

So the impact on aggregate earnings, as tallied by aggregators, was to add roughly 9 to 11 percentage points to the apparent Q1 2026 growth rate, depending on the snapshot used. In plain terms, the aggregators’ headline numbers were not wrong, but they were meaningfully distorted by a small number of very large non-operating investment gains.

***

Conclusions:

1. LLMs are not created equal and do not evolve equally. They sometimes can be downright wrong and misleading.
2. Like always in finance, who you read or talk to matters a lot … Don’t rely on one single source.
3. Be careful with 2026 and 2027 earnings and margins data. Analysts and strategists are not all thorough and precise.
4. Goldman Sachs, while pointing out the significant non-operating profits of Q1, now blends itself with the consensus camp and omits “normalizing” S&P 500 profits in its more recent reports.

GS On May 1:

The hyperscalers’ earnings growth this quarter was boosted by an unusually large contribution from equity stakes in private companies. Alphabet and Amazon generated “other income” totaling $53 billion in Q1 2026, which accounted for nearly 60% of those two companies’ income in Q1 and 34% of the total $155 billion in income this quarter across the five largest hyperscalers. This represents the group’s largest collective share of earnings attributable to “other income” in at least a decade. Of this $53 billion in “other income,” $49 billion was explicitly due to equity stakes in private companies.

Combining realized results and consensus estimates, S&P 500 year/year EPS growth is tracking at 25%, more than twice the consensus estimate of 12% coming into the season. Much of this strength is attributable to the mega-cap technology stocks, some of whose earnings have been boosted by the appreciation of equity stakes in private companies. However, even excluding those figures, S&P 500 EPS growth would be on pace to register 16% in Q1, the strongest quarterly growth rate since 2021.

On June 12, GS official EPS for 2026 is set at $340, up 24% YoY, slowing to +13% in 2027. And that is before upcoming earnings results which will need to mark-to-market the continued appreciation in private AI companies during Q2 and Q3. Particularly GOOG, NVDA and CSCO.

Rough calculations suggest that 2026 S&P 500 EPS are currently inflated by about $20 and its P/E ratio by about 1.3x.

Ed Yardeni:

Analysts’ consensus expected long-term earnings growth (LTEG) jumped again to 24.0% at an annual rate for the next five years during the week of June 12. That’s a record high and twice as fast as the average of this series since 1985. It’s also as unlikely to be achieved as colonizing Mars.

The forward earnings of both the S&P 400 MidCap and S&P 600 SmallCap have been rocketing to new highs, along with that of the S&P 500, in recent weeks

Goldman’s numbers don’t seem to reflect Yardeni’s charts, do they?

S&P 500 median earnings rose 14% in Q1 but are expected to return to their 9-10% historical growth:

But who cares about earnings?

Big Mo in Action (Goldman Sachs)

Volatile market rotations during the past week continued to follow the pattern of the sharpest Momentum rallies in recent decades.

Before the sell-off that started last week, a narrow-breadth Momentum rally had driven the sharpest two-month S&P 500 return since 1971, when scaled relative to volatility.

Following similarly sharp rallies in the past, the Momentum factor has usually struggled during the subsequent few months.

In addition to the historical precedent, the current market characteristics of elevated trading leverage, still-narrow market breadth, and uncertainty about both the macro outlook and the AI build-out suggest volatility will persist.

GROWTH VS VALUE IN THE REAL AI WORLD

The most important AI leaderboard today isn’t benchmarks. It’s adoption.

OpenRouter’s rankings track real-world usage across millions of developer requests, revealing which models people actually choose in production—not just which models top synthetic tests.

#OpenRouter‘s rankings show a striking trend: Chinese models are rapidly overtaking US models in real-world developer usage. Models from DeepSeek, Kimi, MiniMax, GLM, and Qwen now account for a large share of token consumption across the platform.

Why?  Because AI is becoming an economics game.

For many workloads, DeepSeek delivers strong performance at a fraction of the cost of leading US models. When developers are processing billions of tokens, cost-performance wins.

The AI race is no longer just about who builds the smartest model. It’s about who delivers the best value. Right now, DeepSeek looks like one of the biggest winners. (@C_Barraud)

Rankings at openrouter.ai/rankings.

  Here’s someone who has real value for someone:

As Trump’s top personal attorney, Epshteyn has overseen a wave of unprecedented civil litigation from a president against the news media and social media companies.

• Trump has won or forced settlements with ABC ($16 million), CBS ($16 million, plus public service announcements), Meta ($25 million), Google ($22 million) and X ($10 million).
• Other Trump media targets still in litigation: BBC, CNN, The New York Times, The Wall Street Journal and even the Pulitzer Committee. (Axios)

Deal or no deal?

The Guardian has details other media don’t:

“Iran hasn’t reached a final conclusion about the agreement,” Foreign Ministry spokesman Esmail Baghaei said, according to state media. “We will announce it when we reach a conclusion.”

The Iranian foreign ministry spokesperson, Esmail Baghaei, said large parts of the text under negotiation had been finalised but Iran would not compromise on its red lines

Tasnim, the semi-official Iranian news agency, wrote that “until a potential understanding is announced by Iran, any news from Trump on this matter should be dismissed”.

A diplomat briefed on the talks said that the deal had largely been agreed to several weeks ago but that there was still a “50% chance” that it will collapse. “There are a lot of potential spoilers,” the diplomat said.

Financial markets believe that Trump’s 39th announcement (per CNN) of an imminent deal since March 23 will be his last.

A deal now would be timely. Read below.

BTW, in today’s FT:

A regime insider said the war had fundamentally reinforced the leadership’s belief that military power, rather than diplomacy, remains the ultimate guarantor of security.

“More than ever before, we have concluded that building trust is a meaningless strategy,” he said.

“Only strength can serve as a deterrent, not arguments in international forums about our rights. The enemy must be convinced of our capabilities and must never be allowed to miscalculate again. Iran is demonstrating in practice that it is prepared to go further than its adversaries.”

CONSUMER WATCH

From the Bank of America Institute:

Consumer spending remains resilient, with little sign that higher gas prices have dented momentum yet. In May, Bank of America total aggregated credit and debit card spending per household increased 5.1% year-over-year – the largest gain in nearly four years – following the 4.8% YoY rise in April.

While higher YoY gasoline spending is contributing to this strength, total spending excluding gas still rose 3.9% YoY in May, only a small easing from 4.0% YoY in April.

Seasonally-adjusted (SA) spending per household rose 0.1% month-over-month (MoM) in May, following a 0.5% increase in April. This follows gains of 0.6% and 0.9% MoM in the prior two months.

If you stopped reading after BofA’s first 4 words, or after the +5.1% YoY growth rate, you are missing the reality.

• May’s YoY data is against the worst month last year:

• The seasonally adjusted monthly trend was weakening since February but crashed in May at +0.1% in nominal dollars. This is cum-gas. BofA does not provide the MoM ex-gas data but it is clearly down in May and almost flat since February, again in nominal terms.
• My proxy for retail sales inflation (0.35x CPI-Durables + 0.65x CPI-Nondurables) is up 0.9% MoM in May after +1.1% in April and 2.1% in May. Applied to BofA SA data, we get real sales down -0.8% in May after -0.6% in April. Last 2 months annualized: –8.7%.
• On a YoY basis, retail inflation was +6.4% in May after +5.3% in April and +4.0% in March. Real sales are thus down -1.3% YoY in May after -0.5% in April.

BofA’s narrative adds: “Beneath the remarkably robust overall spending picture, are there any signs of weakness in consumer behavior? Largely, the answer appears to be “no.””

Remarkable indeed!

The bank adds that “higher-income households saw stronger growth of 5.4% YoY”.

With my inflation proxy, that is down 1.0% YoY. So even the wealthiest are retrenching in real terms.

After the May 14 release of the official retail sales data, I wrote that real sales were down 0.5% MoM in March and down 0.6% in April, in line with BofA’s card data using my inflation proxy.

On May 25, Walmart released its Q1’26 sales ending in April (all YoY):

• Comp sales ex fuel:      +4.1% vs +4.5% in Q4’25
• Transactions ex fuel:    +3.0% vs +1.6% in Q4’25
• Average ticket ex fuel: +1.1% vs +2.8% in Q4’25

The US largest retailer’s sales were slowing with average ticket up only 1.1% when management said that “like-for-like inflation was +4.4% ex-fuel in Q1”. When WMT released its Q1’25 average ticket growth of 2.8% it said that inflation was “effectively flat or slightly deflationary”.

Notice the huge change in real growth rates? The large decline in real sales?

Per BofA data, the weakness got worse in May.

These are retail sales, essentially goods consumption. Hopefully, services will sustain total consumer spending. Note also that CPI inflation can differ from PCE inflation (different weights, substitutions).

But the recent trends in real demand are not good and certainly not in the mainstream narratives.

Obviously, lower oil prices post a possible deal with Iran would help and provide hope that there has not been too much damage already.

Yesterday, the Atlanta Fed reported that its wage growth tracker was +3.5% in May, down from +4.3% one year ago.

Meanwhile, CPI inflation accelerated from +2.4% to +4.2%.

US Producer Prices Rise at Fastest Pace Since November 2022

The producer price index increased 6.5% from a year earlier, the most since November 2022, according to Bureau of Labor Statistics data out Thursday. It advanced 1.1% from April.

A core measure of prices that excludes food and energy increased 4.9% from a year earlier. (…)

“Most of the increases were directly tied to higher costs for fuels and natural gas, but goods prices outside of energy jumped in May as supply chain stress bleeds into other materials and components,” Ben Ayers, a senior economist at Nationwide, said in a note.

“With fuel prices fading so far in June, this may be the peak for producer price inflation but the aftereffects for consumer prices are likely to linger over the remainder of 2026,” he said. (…)

Food prices, meanwhile, rose 0.6%, the most in three months. Grocery costs have been moving higher thanks to a combination of factors including bad weather, the war and tariffs. Fertilizer materials costs were up 28% from a year earlier.

A measure of inflation pressures earlier in the production process — the cost of processed goods for intermediate demand excluding food and energy — rose from April by the most since 2021. Plastic resins and materials, a key input for a vast array of consumer goods, surged 14% on the month.

Several components of the PPI are also of particular interest because they feed into the the Fed’s preferred inflation gauge, the personal consumption expenditures price index. (…)

“May’s PPI details suggest upstream cost pressures are building. Goods categories are growing much slower than industrial categories, meaning costs haven’t yet reached the consumer — potentially pointing to continued price sensitivity by shoppers. Elsewhere, higher transportation costs, driven by fuel and capacity constraints, are rising, as are inputs into production processes.”

The report also provided updates on two other emerging sources of inflation pressures: data centers and defense production. The prices of electronic components and accessories fell from April for the first time in more than a year, but were still up nearly 27% from May 2025.

At the same time, prices associated with government purchases for defense were up almost 15% from a year earlier. Census Bureau figures published last month showed defense-related capital goods orders surged in April to the second-highest level on record, with economists citing replenishment of munitions destroyed in the Iran war as one possible driver.

Also:

PPI goods’ final demand surged 2.8% MoM, the largest one-month advance on record, lifting the YoY rate to 10.4%, the highest since October 2022.

PPI services’ final demand inflation remained high at 4.9% YoY.

PPI ex-food, energy, and trade services was +0.8% MoM, +5.1% YoY.

Goldman Sachs: “The PPI components relevant for PCE were slightly stronger than our expectations. Based on details in the CPI and PPI reports, we estimate that the core PCE price index rose 0.31% in May (vs. our expectation of 0.28% before the PPI release), corresponding to a year-over-year rate of +3.38%.”

Many pundits rejoiced seeing that Wednesday’s CPI-Goods was somewhat muted. This Bloomberg piece explains the lags:

A gauge of wholesale prices of plastic resins and materials jumped by 14% to an almost four-year high last month as the Iran war choked off supplies of key components. For producers like Shawn Gross, whose Corry, Pennsylvania-based company supplies molded parts used in automotive and heating systems, the squeeze is reaching a breaking point.

“We are going to need to pursue the price increases with our customers more aggressively,” said Gross, chief executive officer of Viking Plastics. “What the world needs to understand is there is going to be a real impact that is not even yet felt.”

Plastics are everywhere — from snack packaging to refrigerator parts — and about 98% of it is made of fossil fuels. One reason why US consumers haven’t felt the full impact of the war-driven price increases in petrochemicals like polyethylene that are used in everyday products is because costs move slowly through supply chains.

Gross, whose suppliers include Dow Inc. and clients include Ford Motor Co., is at the center of it all. Prices for some polyethylene inputs used by Viking have risen more than 40% this year, he said.

Producers like Dow have already pushed through price hikes to businesses that convert plastic resins into packaging and other products. As those costs are passed from one company to the next, higher prices ultimately make their way to consumers.

“You need to get the dollars and cents out of the customer, and that’s going to trickle into the price of vehicles being higher,” Gross said. “It’s going to be every industry.” (…)

Some buyers are receiving only about 70% of the resin they need, he said, while lead times for certain materials have stretched from one to three months. Fresh-food packaging — for products like produce, bakery goods, meat and cheese — will get hit first.

If shortages persist, shoppers could begin seeing gaps on shelves by August and September, he said.

“We are anticipating further inflation in a number of non-food categories as higher resin costs start to flow into cost-of-goods,” Chief Financial Officer Gary Millerchip said during the retailer’s quarterly earnings call. “Particularly if oil prices remain at elevated levels, it’s likely to see some increases in items that have sort of plastic components or polyester or cotton because of the impact of higher resin costs.”

Inflation you say?

• Amazon 1997 IPO: 2.9x revenues.
• Google 2004 IPO: 8.5x revenues.
• Facebook (now Meta) 2012: 17.9x revenues
• SpaceX 2026 IPO: 93.6x revenues.

Trump administration delays opening of Gordie Howe bridge on eve of ribbon cutting

The opening of the Gordie Howe International Bridge between Windsor, Ont., and Detroit was postponed indefinitely at the behest of the Trump administration, one U.S. and one Canadian industry source said, just a day before the long-delayed span’s planned Friday ribbon cutting.

The U.S. industry source said the American government slammed the brakes on the opening because Commerce Secretary Howard Lutnick and Pete Hoekstra, Washington’s ambassador to Canada, want to first negotiate a deal to help Michigan’s Moroun family, the billionaire owners of the existing Ambassador Bridge, mitigate their losses from competition by the new, publicly owned bridge.

The Globe and Mail agreed not to name the sources because they were not authorized to disclose the information.

The Canadian federal government agreed to pay the full $6.4-billion price tag for the structure in 2012 after the Michigan legislature refused to chip in under lobbying from the Morouns. Ottawa will be repaid with toll revenue from the bridge, which is jointly owned by Canada and Michigan.

Earlier this year, U.S. President Donald Trump threatened to stop the bridge from opening and said Canada should give the U.S. government “at least one half” of the asset. Mr. Trump’s threat came shortly after Matthew Moroun, chairman of the company that owns the Ambassador Bridge, donated US$1-million to a pro-Trump campaign group and reportedly met with Mr. Lutnick.

The Gordie Howe is meant to speed traffic over the busy international crossing and trade route, providing direct access between highways – avoiding the current route through Windsor streets on the approach to the Ambassador, which opened in 1929 – and offering more extensive cargo screening facilities.

The U.S. industry source said the opening of the Gordie Howe was scheduled after Michigan Governor Gretchen Whitmer spoke with Susie Wiles, Mr. Trump’s chief of staff, and believed she had the White House’s sign-off. Mr. Lutnick, who learned of the planned ribbon-cutting on Monday night while attending a New York Knicks game with Mr. Trump at Madison Square Garden, was caught off-guard by the news, the source said.

The source, who was briefed by the Trump administration, said someone from the administration contacted Canadian Intergovernmental Affairs Minister Dominic LeBlanc to demand the bridge not open.

Mr. Lutnick and Mr. Hoekstra have been working on a deal with the Morouns to ensure the Ambassador doesn’t lose all of its toll revenue to the Gordie Howe, the person said, and they want to conclude the agreement before the rival bridge opens. Some potential ways to do this would be to fix tolls at particular levels to avoid Gordie Howe outcompeting the Ambassador, or divert specific types of traffic – like Nexus travellers, for instance – to the Ambassador, the source said. (…)

Ms. Wright, who was en route to El Paso, Tex., has driven trucks over the Ambassador Bridge, sometimes up to three times a week, for 26 years. She called the 97-year-old span a “piece of junk.” The duty free road is littered with pot holes that damage her vehicle. Most weeks she spends hours in delays. (…)

Inderjeet Singh, a trucker taking a brief break at a park bench outside the Esso, said he waited over eight hours at the Ambassador Bridge on Wednesday. There was a problem with the load, and the lineup for inspections was backed up.

Mr. Singh also wants to pay cheaper rates to cross into Detroit. The Gordie Howe has set its tolls at about half those of the Ambassador. (…)

Mr. Trump first took an interest in the Gordie Howe in February, complaining on his Truth Social platform that Canada owns the bridge and will collect its toll revenue, even though Ottawa paid to build the span and the ownership is shared with Michigan.

The New York Times, citing unnamed government officials, reported that Mr. Moroun had lobbied Mr. Lutnick just hours before Mr. Trump’s post. Campaign finance disclosures show that Mr. Moroun gave US$1-million to the MAGA Inc. political action committee the previous month. (…)

How often do we read White House spoke persons tell us that

• “he is an unrelenting warrior fighting exclusively for the forgotten men and women of America against entrenched political interests”
• “the first duty of the American government is to protect American citizens”
• “he is fighting to ensure the federal system serves the public rather than self-interested elites”
• “a president who refuses to back down to special interest groups or congressional obstruction”