TheChek News Flow

EQUITY STRATEGY | SPX '26 EPS vs. Multiple Divergence — The Mag7 Confession Season Begins | Q1 2026 Earnings Preview

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

I. THE CORE DISLOCATION: EPS STILL AT $323, MARKET PRICING RECESSION

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

The BofA chart making the rounds this morning crystallises what is arguably the cleanest macro signal of the cycle: S&P 500 2026E EPS has been revised UP +4% since January 1st to $323 — almost entirely on Mag7 margin and capex monetisation assumptions — while the index itself has violently de-rated. The divergence is not noise. It is a direct read on multiple compression: the market is not buying the $323 story.

The implied forward P/E on the way down is now ~18.8x vs. a 22x+ peak six weeks ago. Historically, that compression happens either because (a) estimates are wrong, or (b) the macro trajectory deteriorates faster than analysts can revise. Right now, both risks are live simultaneously.

What is not yet priced: analyst estimates are a lagging variable. Buy-side knows this. The sell-side coverage universe has NOT moved the Mag7 EPS pencil materially — cuts are coming on guidance, not on forward assumptions. That is the trap embedded in the BofA chart. The moment guidance wobbles on the April/May reporting cycle, the $323 floor becomes the $295–$305 ceiling, and re-rating lower accelerates.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

II. THE FULL Q1 2026 EARNINGS CALENDAR — SUPPLIERS FIRST, MAG7 AFTER

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

⚡ TIER 1 — CRITICAL READ-THROUGHS BEFORE MAG7 REPORTS

┌─────────────────────────────────────────────────────────────────────────────┐

│ DATE │ COMPANY │ TICKER │ WHAT TO WATCH │

├─────────────────────────────────────────────────────────────────────────────┤

│ Apr 15 (BMO) │ ASML │ ASML │ EUV order book, China normalisation │

│ Apr 16 (BMO) │ TSMC │ TSM │ Hyperscaler CoWoS demand, 2nm ramp │

│ Apr 16 (AMC) │ Netflix │ NFLX │ Ad spend environment, digital CPM │

│ Apr 22 (TBC) │ SAP SE │ SAP │ Enterprise IT budget read-through │

└─────────────────────────────────────────────────────────────────────────────┘

⚡ TIER 2 — MAG7 REPORTING WINDOW (APRIL 28 – MAY 20)

┌─────────────────────────────────────────────────────────────────────────────┐

│ DATE │ COMPANY │ TICKER │ CONSENSUS EPS │ CONSENSUS REV │ KEY │

├─────────────────────────────────────────────────────────────────────────────┤

│ Apr 28 (AMC) │ Tesla │ TSLA │ ~$0.50 │ ~$23–24B │ AUTO │

│ Apr 28 (AMC) │ Alphabet │ GOOGL │ ~$2.10–2.20 │ ~$89–91B │ CLOUD │

│ Apr 29 (AMC) │ Amazon │ AMZN │ ~$1.95–2.10 │ ~$155–157B │ AWS │

│ Apr 29 (AMC) │ Meta │ META │ ~$5.25–5.50 │ ~$41–43B │ AI/AD │

│ Apr 30 (AMC) │ Apple │ AAPL │ ~$1.60–1.65 │ ~$94–96B │ IPHONE│

│ May 7 (AMC) │ Microsoft │ MSFT │ ~$3.45–3.55 │ ~$73–75B │ AZURE │

│ May 20 (AMC) │ Nvidia │ NVDA │ ~$0.90–0.95 │ ~$43–45B │ DC/AI │

└─────────────────────────────────────────────────────────────────────────────┘

Note: Consensus figures as of April 5, 2026. Dates subject to confirmation. Alphabet and Amazon dates remain unconfirmed; Tesla confirmed Apr 28.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

III. DEEP DIVE — MAG7 STOCK BY STOCK

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

───────────────────────────────────

1. NVIDIA (NVDA) | Reports ~May 20

───────────────────────────────────

Last print (Q4 FY2026, Feb 25): Revenue $68.1B (+73% YoY, +20% QoQ). Data Centre alone $62B (+75%). Non-GAAP EPS $1.62 vs. $1.54 est. Gross margins held at 75%. Q1 FY2027 guidance: $78B (±2%) — a further ~15% sequential step-up.

Management guided that Vera Rubin next-gen AI chips are tracking SIX MONTHS AHEAD OF SCHEDULE — one of the most significant positive datapoints in the tech cycle right now.

Key risk into Q1 FY2027: Blackwell ramp deceleration, gross margin pressure from CoWoS packaging constraints (TSMC-dependent), and export control overhang on H20 series into China. Any guide-down from the $78B level will be interpreted as a demand air pocket — and will re-price the entire AI infrastructure trade.

What we want to hear: (i) Vera Rubin pull-forward orders from hyperscalers, (ii) inference chip ramp (GB200 NVL72 rack systems), (iii) no margin compression below 73%.

Bull case: $85B+ Q2 guide. Bear case: $72–74B with margin pressure — stock gaps -15% or more.

───────────────────────────────────

2. MICROSOFT (MSFT) | Reports ~May 7

───────────────────────────────────

Last print (Q2 FY2026, Jan 29): Revenue $81.3B (+17% YoY). Azure +39% YoY — the standout metric. Microsoft Cloud $51.5B (+26%). Commercial RPO surged to $625B (+110% YoY) — a staggering forward demand indicator. GAAP EPS $5.16 (+60%). Capex guided at $99B for FY2026 (June year-end), with CFO Amy Hood confirming further growth in FY2027. OpenAI relationship evolving — Microsoft is simultaneously diversifying toward Anthropic ($5B investment).

The Azure +39% number is the single most important data point in the Mag7 universe. If it holds or accelerates in Q3 FY2026 (the quarter being reported in May), the AI monetisation thesis is intact. If it decelerates to +32–34%, expect material estimate cuts across the entire hyperscaler complex.

Key watch: Azure growth rate vs. expectations, Copilot seat monetisation velocity, capex commentary for FY2027.

───────────────────────────────────

3. META PLATFORMS (META) | Reports Apr 29

───────────────────────────────────

Last print (Q4 2025, Jan 29): Revenue $59.9B (+24% YoY). FY2025 revenue $201B (+22%). EPS $8.88 vs. $8.21 est. (+8.1% surprise). Ad impressions +18% YoY; average price per ad +6%. DAP 3.58B (+7%). FY2025 capex: confirmed in the $60–65B range — above prior consensus of $50.7B.

2026 capex guidance: Zuckerberg guided $110–125B+ for 2026 AI infrastructure. This is the single largest capex commitment of any company globally as a % of revenue. The market initially sold this hard (-28% from peak). Since then, META has partially recovered on ad monetisation resilience.

Q1 2026 key read: Ad revenue growth (is the +20% YoY revenue trajectory holding?), Reality Labs losses vs. expectations, any revision to the $110–125B capex band. If Zuckerberg signals capex optimisation, read that as ROI pressure — not discipline. If he doubles down, the market will want evidence of Llama/AI monetisation timelines.

───────────────────────────────────

4. ALPHABET (GOOGL) | Reports ~Apr 28

───────────────────────────────────

Last print (Q4 2025, Feb 4): EPS $2.82 vs. $2.63 est. Revenue $113.83B vs. $111.43B est. Net income +30% YoY to $34.46B. Google Cloud $17.66B (+48% YoY — the standout). YouTube ad revenue $11.38B — missed $11.84B expected. 2026 capex guidance: $175–185B — more than DOUBLE 2025 levels. Alphabet is betting the entire company on AI infrastructure.

Q1 2026 concern: YouTube miss in Q4 raises the question of whether digital ad CPM is starting to soften at the margin — a critical read-through for the broader ad market. Google Cloud growth of +48% must continue: if it decelerates to +35–38%, the AI monetisation premium evaporates. Search ad market share is also under scrutiny as AI-native search (Perplexity, OpenAI SearchGPT) gains enterprise adoption.

Key watch: Cloud revenue growth rate, Search ad market share commentary, YouTube CPM trends, any revision to the $175–185B capex commitment.

───────────────────────────────────

5. AMAZON (AMZN) | Reports ~Apr 29

───────────────────────────────────

Last print (Q4 2025): EPS $1.95 — slight miss vs. $1.97 consensus. AWS: consensus was $29.4B for Q1, operating margin ~35%. FY2026 capex: guided to $105B — nearly double 2023 levels.

AWS is the most important cloud read-through for enterprise AI spending. AWS growth vs. Azure growth will define the Q2 sector narrative. Amazon has additional complexity: tariff exposure on the e-commerce side creates a dual-risk print — if the core retail margin is squeezed while AWS is also guiding cautiously, the stock re-rates to a new lower bound. FX is an additional headwind given USD strength.

Key watch: AWS revenue growth rate (consensus ~$29.4B for Q1), operating margin trajectory, any tariff impact disclosure on retail COGS, North America advertising revenue (a stealth AI monetisation engine).

───────────────────────────────────

6. APPLE (AAPL) | Reports Apr 30

───────────────────────────────────

Last print (Q1 FY2026, Jan 29): EPS $2.84 vs. $2.67 est. Revenue $143.8B (+16% YoY). iPhone +23% to $85.27B. China recovery: +38% growth to $25.53B — the most positive surprise in years. Services $26.34B (+14%). Mac -7%. Wearables -2%. Demand outpaced supply.

Apple is paradoxically the SAFEST of the Mag7 into earnings — the last print was strong, iPhone demand held, and the Google Gemini deal for Siri overhaul adds an AI monetisation optionality layer that has not yet been priced by most models.

Key risk into Q2 FY2026 (April 30 report, which covers the January–March quarter): tariff exposure on Chinese manufacturing — Apple has ~90% of iPhone production in China. Every 10% tariff on Chinese goods is estimated at a ~$900 gross profit hit per iPhone. Tim Cook has been actively building out India capacity, but the ramp cannot cover FY2026 volumes. This is the one name where the tariff narrative could genuinely move the needle on estimates.

Key watch: Gross margin guidance (tariff pass-through), China revenue (can the +38% trajectory hold?), Services growth, India production ramp timeline.

───────────────────────────────────

7. TESLA (TSLA) | Reports Apr 28

───────────────────────────────────

Last print (Q4 2025, Jan 28): Revenue $24.9B (-3% YoY) — first annual revenue decline in company history. FY2025 revenue $94.83B vs. $97.69B in 2024. GAAP net income Q4 $840M (-61%). Non-GAAP EPS $0.50 vs. $0.40 est. (+25% surprise). Gross margin 20.1%. Cash $44.1B. Deliveries Q4: 418,227 vehicles.

Tesla is structurally the most challenged name in the Mag7 complex. Top-line has stalled, BYD has overtaken as global EV leader, and the valuation at 200x+ forward earnings is entirely dependent on Optimus/Robotaxi optionality — neither of which is contributing to 2026 earnings.

Elon's political visibility has created measurable brand damage in Europe and coastal US — a real headwind that has not yet been fully modelled into delivery forecasts. Q1 2026 delivery data (expected shortly) will be the first read on whether the Cybercab ramp is offsetting Model S/X weakness. Any delivery miss below 400K will re-open the debate on the 200x multiple.

Key watch: Q1 deliveries vs. 400K threshold, gross margin resilience, Optimus/FSD timeline update, any Robotaxi regulatory news.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

IV. THE CRITICAL SUPPLIER READ-THROUGHS — TRADE THESE BEFORE MAG7

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

───────────────────────────────────

A. TSMC (TSM) | Apr 16 — THE MOST IMPORTANT REPORT OF THE SEASON

───────────────────────────────────

TSMC guided Q1 2026 revenue of $34.6–35.8B (+38% YoY). Gross margin 63–65%. FY2026 revenue expected to grow ~30% in USD terms. The company is ramping 2nm Gate-All-Around (GAA) technology and managing Arizona Fab 3 costs simultaneously.

Why it matters: TSMC is the gating variable for the entire AI stack. NVIDIA relies on TSMC for 100% of its H300 and B200 accelerators. A TSMC beat and confident call validates hyperscaler demand is NOT rolling over. A miss or cautious commentary — particularly on HPC (High Performance Computing) order visibility — is an immediate negative read-through for NVDA, ASML, and by extension every hyperscaler capex story.

Additional risk: The Helium Crisis in late February (Ras Laffan, Qatar disruption) raised output questions. Watch for any supply-chain commentary on fab productivity.

● TSMC beat + $36B+ rev + strong HPC commentary → Buy Nvidia, Buy Microsoft, Buy Alphabet

● TSMC miss or cautious HPC guide → Sell everything in the AI stack before Mag7 reports

───────────────────────────────────

B. ASML (ASML) | Apr 15 — EUV ORDERS ARE THE LEADING INDICATOR

───────────────────────────────────

ASML reports April 15 (BMO). Q1 2026 EPS consensus: $7.67. Revenue guidance: €8.2–8.9B. FY2026 guidance: €34–39B. ASML guided that NVIDIA wafer requirements are rising from ~2.5 wafers to ~10 wafers per product by 2027 — a seismic demand signal for lithography intensity.

Key watch: Net bookings figure (Q4 2025 was a record €13.2B). Any deceleration in bookings = demand pull-forward risk. China: management guided China normalises to ~20% of total sales in 2026. Any upside in China is a risk factor given export control evolution. EUV/High-NA ramp commentary is the forward read on TSMC's 2nm/1.4nm timeline.

● ASML strong bookings + EUV confidence → confirm AI infrastructure capex cycle

● ASML booking slowdown → leading indicator of hyperscaler capex pause — front-run the Mag7 short

───────────────────────────────────

C. NETFLIX (NFLX) | Apr 16 — DIGITAL AD MARKET HEALTH

───────────────────────────────────

Netflix reports April 16 (AMC). Q1 2026 EPS: $0.76 consensus (+15.2% YoY). Revenue: $12.17B. FY2026 revenue guided at $51B (+14% YoY), operating margin 31.5%. Ad sales grew 2.5x in 2025 and management expects a DOUBLING in 2026 to ~$3B.

Netflix's ad-supported tier CPM pricing and ad sales growth is the most direct read on the digital advertising market that reports BEFORE Alphabet and Meta. If Netflix ad revenue is tracking strongly in Q1, that de-risks the Alphabet YouTube miss in Q4 and confirms Meta ad impressions are still monetising efficiently. A Netflix ad softness read = risk-off on both GOOGL and META into earnings.

● Netflix ad acceleration → green light for Alphabet and Meta estimates

● Netflix ad miss → go defensive on digital ad exposure ahead of GOOGL/META

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

V. THE DECISION FRAMEWORK — HOW TO TRADE THE SEQUENCE

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

The earnings season functions as a sequential de-risking (or re-risking) event. The sequencing matters:

STEP 1 — APR 15/16 (ASML + TSMC + Netflix): The true sentiment barometer. These three names tell you everything about the underlying demand picture before the Mag7 CEO conference calls add narrative complexity. Trade the read-throughs.

STEP 2 — APR 28/29/30 (Tesla + Alphabet + Amazon + Meta + Apple): The EPS anchors. Five of seven Mag7 in a three-day window. Expect extreme volatility. Each name's capex commentary will be cross-referenced against the others — any inconsistency in hyperscaler demand language (AWS vs. Azure vs. Google Cloud) will be traded aggressively.

STEP 3 — MAY 7 (Microsoft): Azure growth rate is the definitive read on enterprise AI adoption velocity. This is the last nail in the coffin or the all-clear signal.

STEP 4 — MAY 20 (Nvidia): The season's final boss and the most binary print of the cycle. Guidance vs. the $78B+ Q1 FY2027 implied baseline — any miss here re-opens every debate about AI infrastructure overcapacity.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

VI. THE EPS SCENARIO TABLE — WHERE DOES $323 LAND?

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SCENARIO EPS 2026E FAIR P/E SPX IMPLIED

─────────────────────────────────────────────────────────

Estimates hold $323 20x 6,460 ✅

Mild cut (-8%) $297 19x 5,643 ⚠️

Hard cut (-15%) $275 17x 4,675 ☠️

Stress cut (-22%) $252 15x 3,780 🔴

The stress scenario assumes Mag7 guide-down + multiple compression from macro deterioration. It is not the base case. But at current index levels, it is not fully priced out either.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

VII. OUR POSITIONING VIEW

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

We are NOT buyers of the index ahead of earnings confirmation.

The BofA chart documents a classic late-cycle trap: the multiple compression has been violent, but the EPS anchor has not yet been stress-tested by management guidance. Until the confession season is complete — i.e., after Microsoft and Nvidia report in May — the $323 EPS line remains a working assumption, not a floor.

SELECTIVE TACTICAL VIEWS:

→ OVERWEIGHT: TSMC (TSM) into Apr 16 — strongest risk/reward in the AI stack with confirmed demand guidance

→ OVERWEIGHT: Microsoft (MSFT) — Azure trajectory and RPO backlog provide the most defensible earnings quality

→ UNDERWEIGHT: Tesla (TSLA) — 200x forward multiple with structurally declining deliveries; no margin of safety

→ UNDERWEIGHT: Apple (AAPL) into tariff clarity — China manufacturing exposure not yet quantified

→ NEUTRAL/WATCH: Meta (META) — capex discipline is the single swing variable; Zuckerberg tone on AI ROI timeline will define the stock in H2

→ NEUTRAL: Alphabet (GOOGL) — Cloud recovery is real; YouTube softness and antitrust remain structural overhangs

→ AVOID: Adding Nvidia beta ahead of May 20 print — asymmetric risk to guide-down after $78B implied baseline

The mid-May window — post all seven Mag7 prints — is where we see the highest-conviction entry point for the index if estimates prove durable.

The Elon Musk-led company now looks to test an upgraded version of its giant rocket in May after previously targeting March and April

SpaceX delayed the next test flight of its Starship rocket until early to mid-May, after previously aiming for early March and April.

NASA plans to use a version of Starship in 2028 to land astronauts on the moon, but some officials have been skeptical of its readiness.

The upcoming Starship flight, following 11 previous tests, features numerous upgrades like a redesigned propellant system and improved Raptor engines.

SpaceX delayed the next test flight of its massive Starship rocket, pushing off the highly anticipated mission by at least another month.

SpaceX Chief Executive Elon Musk said Friday the next launch attempt for the roughly 400-foot rocket would occur sometime in early to mid-May. In January, Musk said the new version of Starship would launch in early March, and he later pushed the date to around April 7.

Touted for years by Musk as the key to taking humans to Mars, SpaceX is now looking to show that Starship can work, including by lofting the company’s own satellites into orbit. It is developing the rocket as the company prepares for a highly anticipated public offering that bankers believe will make SpaceX one of the most valuable companies in the world.

Rocket development is always a difficult endeavor, with margins between success and fiery failures relatively thin. SpaceX’s Starship is designed to be the most powerful rocket ever built, consisting of a booster that propels a large spacecraft into space. Both are intended to be rapidly and fully reusable.

The National Aeronautics and Space Administration is also looking to have a version of Starship ready for 2028, when the U.S. space agency plans to use the vehicle to land astronauts on the lunar surface for the first time in over 50 years.

Some current and former spaceflight officials are skeptical that the Starship lander will be ready for that operation, given the amount of development work ahead and the relatively tight time frame. SpaceX has said it offered the agency ideas to simplify the planned landing operation.

Texas-based SpaceX has conducted 11 flight tests of Starship over the last three years, but the next one is expected to be one of the most important to date.

The version of the rocket SpaceX plans to put through its paces on the next flight features numerous upgrades and engineering changes. Those include a redesigned propellant system on its booster and improved Raptor engines.

Country faces assaults from militias in Iraq, Hizbollah and Israel’s military as its fledgling government seeks to sit out conflict

Syria is facing cross-border attacks from Shia militia groups in Iraq and from Israel as it struggles to remain on the sidelines of the US-Israeli war on Iran.

Damascus’s fragile government is seeking instead to position the country as an energy transit hub for nations affected by the shutdown of the Strait of Hormuz as the US signalled the war, now in its fifth week, would run for weeks more.

Earlier this week Syria said it had intercepted a large-scale drone attack on former US bases near the Iraqi border, the latest assault by Shia militia groups. Last month Syrian troops came under artillery assault by the Hizbollah militant group in neighbouring Lebanon, which like the groups in Iraq joined the fray in support of Tehran last month.

Within Syria itself, a shadowy Shia militia group has taken credit for several attacks on US and Israeli assets over the past three weeks. The group was formed earlier this year, vowing to fight Ahmed al-Sharaa’s Sunni government.

“We had enough war,” the president, who seized power in 2024 after years of civil conflict, said at an event hosted by the Chatham House think-tank in London this week. “We are not ready ​for another war experience.”

Israel — which has been using positions it occupies in southern Syria to stage attacks on Lebanon — also launched a wave of air strikes on southern Syria last month following alleged clashes between government security forces and Druze militiamen, threatening to deepen domestic unrest. Israel said, as it has done with previous interventions in Syria, that it was coming to the defence of the Druze minority.

Damascus, meanwhile, has sent thousands of its troops to bolster its borders and limit the flow of weapons and fighters, defence officials have said.

Those movements have fuelled deep concerns in Lebanon and Iraq, where government officials and pro-Iran groups speculated that Syrian forces could launch cross-border incursions. Syrian officials and analysts say that worry is unfounded. 

The US envoy to Syria denied a Reuters report last month that Washington was pressing Damascus to operations inside Lebanon to help disarm Hizbollah. But regional and western officials told the FT serious conversations about the possibility had taken place with US officials over a period of months, both before and during the war.

“Syria’s armed forces don’t have the capacity to do anything like this,” said Kheder Khaddour, a Syria analyst at the Carnegie Middle East Center in Beirut. “They barely have enough forces to cover their own territory.”

Following 14 years of devastating civil war that placed Syria at the epicentre of the Levant’s instability, Damascus now finds its position inverted. Neighbouring countries are increasingly caught up in the fallout from the US-Israeli war against Iran, but Syria has remained relatively insulated from its effects, analysts said, as it tries to piece itself together under its new rulers.

“For over a decade, the country served as the central arena where regional and international rivalries played out. Today, by contrast, it has repositioned itself as a neutral actor,” wrote Haid Haid, Syria researcher at Chatham House.

That transition process since Sharaa took power in December 2024 has been marred by instability and sporadic violence. But he has worked to reposition Syria away from its former pariah status on the world stage.

Sharaa has deepened relationships with Arab nations — including Lebanon and Iraq, where residual tensions remain from his past ties to al-Qaeda — as well as with Washington and European states. 

Asked whether Syria would remain neutral on the Iran conflict, he said this week: “We want Syria to have ideal relationships with the entire region, with Lebanon, Iraq, Turkey, Saudi Arabia and world powers like the UK, France, Germany and the US.

“I think that Syria is qualified to start a strategic relationship network.”

One country with which Damascus does not maintain formal diplomatic ties is Iran, an ally of former dictator Bashar al-Assad, whom Sharaa ousted in 2024. 

“Unless Syria is targeted by any party, Syria will remain outside any conflict,” the Syrian president said, but he added: “We do not want Syria to be an arena of war. But unfortunately, today, things are not governed by wise minds. The situation is volatile and random.”

Sharaa has also sought opportunity in the chaos, positioning his country as a conduit for energy and telecommunications between Europe, the Gulf and Asia. 

In Germany this week, Sharaa pitched his country and its land routes as a safe alternative to counteract disruptions to supply chains and energy flows from the near closure of the Strait of Hormuz.

“Syria can serve as a safe haven. It can ensure the security of supply chains through its strategic location, as well as energy supplies via the Mediterranean coast,” he told the heads of German companies in Berlin.

“Any disruption in the Red Sea or the Strait of Hormuz further underscores Syria’s potential role as a secure alternative route.”

It follows a yearlong effort to recruit significant investment in the energy sector in Syria, backed by Washington, in which Damascus secured billions of dollars in investment to help revive the ailing sector. 

This week, Iraq began overland fuel exports using Syrian land routes, as the Strait’s shutdown strangles its exports. Baghdad said volumes would expand gradually. But they face constraints from poor infrastructure — which has been damaged by a decade of war and neglect — and security.

As Iraqi militias continue to attack and threaten Syria, there are concerns it could be drawn further into the conflict.

“How long can Syria stay neutral?” Khaddour said. “The longer this war goes on, the more this conflict expands, the greater the risk it could spill over into Syria.”

New rules aim to provide safety and security for tenants, but landlords are anxious

On TikTok, property influencers lament the end of the buy-to-let landlord. “What used to work back in the day is now not just dead but a complete catastrophe,” says Samuel Leeds, a TikTok content creator who focuses on property and tax issues. “This is terrible for most hard-working British landlords.”

The Renters’ Rights Act, set to come into force in May, is one of the most significant changes the residential property market has seen in decades, and many are anxious about its impact.

Since the early days of the legislation, landlords have warned it would lead to an exodus of small-scale property owners, who make up 83 per cent of the UK private rental sector.

While there are some signs of fewer landlords entering the market, the sell-off among landlords appears to have been tempered by falling residential property prices, particularly in London.

“I think the people that were going to exit have already exited,” said Chris Norris, chief policy officer at the National Residential Landlords Association.

Many in the property sector suggest the act will force the buy-to-let landlords who remain to become more professional.

“Landlords with one or two properties, particularly those who entered the sector accidentally through inheritance, relationship change or an inability to sell, are more exposed to rising complexity and costs,” said Louisa Sedgwick, managing director of mortgages at Paragon Bank.

The government argues that the Renters’ Rights Act is a long-overdue effort to provide safety and security to tenants.

Angela Rayner, the former deputy prime minister who helped push the bill through parliament last year, told the FT she “doesn’t apologise” for the new rules and denied it was piling pressure on landlords.

“There are landlords out there that can make a profit without it being at the expense of other people’s safety and security,” she said.

For tenants, the act will mean they can challenge any “unreasonable” rent increases at a tribunal. It will also end bidding wars, by forbidding landlords and letting agencies from soliciting multiple offers on a property. Among the most significant changes will be the abolition of Section 21, which allows landlords to evict tenants without cause.

Landlords are concerned that this will make it more difficult and costly to deal with problem tenants, and property lawyers are reporting a rise in eviction notices before the act comes into force.

“We’ve seen a massive increase in serving the Section 21s before the deadline day,” said Paul Shamplina, founder of Landlord Action, a company that specialises in housing law. 

Shamplina said the firm had seen a 62 per cent increase in the number of eviction instructions year on year, equivalent to “hundreds” of additional notices. Ministry of Justice figures for evictions this year have not yet been released.

From May, landlords will only be able to evict tenants if a reason is given, such as non-payment of rent, antisocial behaviour or because the landlord is moving into the property.

This change is expected to lead to more property tribunal cases and added pressure on an already strained court system; the Ministry of Justice said it was working “to strengthen capacity”, including hiring more judges.

Kristine Ng, a partner at Morr & Co solicitors, said she was seeing more landlord-tenant disputes and evictions as property owners moved to “sell, redevelop or re-let properties on different terms” before the new regulations took effect.

Landlords who do not comply with the act will face fines, including £7,000 if they fail to send a Renters’ Rights Act Information Sheet to tenants, as well as additional penalties for failing to deal with mould and electrical risks.

However, Shamplina argues that tenants are most likely to be the worst hit as the regulations cause landlords to reprice risk and increase rents.

Anastasia Karaseva moved with her daughter to Richmond, south-west London, five years ago. But the former paralegal, who is undergoing treatment for breast cancer, said she was “shocked” when she received a Section 21 notice from her landlord last month.

Karaseva, who is herself a landlord through a property she owns overseas, is working with the Acorn tenants union to fight the eviction.

“It’s just outrageous because they’re doing exactly what the act will prevent them from doing,” Karaseva said.

Her landlord, the Poppy Factory veterans’ charity, told the FT it was evicting “a small number” of tenants to bring properties up to market rate before the new rules on rental increases came into effect.

“The Renters’ Rights Act has meant we have had to move more quickly than we would have liked,” a Poppy Factory spokesperson said, adding that no veterans or charity beneficiaries were affected by the move.

Some landlords have opted to move outside of the mainstream private rental sector. London-based specialist property agency Elliot Leigh said it had seen an influx of landlords interested in multiyear “guaranteed rent” deals, which involve properties being signed over to councils for use as temporary accommodation or social housing.

Last month, the FT reported that Criterion Capital, which provides both private rental and temporary accommodation, had served Section 21 notices to private residents in 130 units across its property portfolio.

“There are growing signs that some landlords are evicting tenants, only to rehouse homeless families at higher rents,” said Susie Dye of the Trust for London, a non-profit group.

Pharma just ‘scratching the surface’ of weight-loss drug market

Novo Nordisk chief estimates industry reaching only 15% of potential customers at most

Pharma companies are just “scratching the surface” of the huge potential market for weight-loss drugs and should focus more on widening access, according to the boss of the company behind Wegovy and Ozempic.

Mike Doustdar, chief executive of Novo Nordisk, told the FT in an interview, “You need to expand the market especially if you work in a business like ours. In the US, there are 110mn people [with] obesity.”

He added that Novo, its main rival Eli Lilly, and US “compounding” pharmacies that make copycat obesity drugs combined “are scratching the surface — we’re getting no more than 10 to 15 per cent of that population at this point in time. There’s very little talk about the rest.”

Doustdar took over as Novo chief executive last August to try to improve the Danish company’s fortunes after a difficult period when it lost its market lead to Lilly and had some disappointing trial results.

It is hoping that the launch of its weight-loss drug Wegovy in pill form this year will help it win back ground, though Lilly has also just won US regulatory approval for its weight-loss pill Foundayo.

Doustdar said the need to expand access was one reason Novo agreed to cut prices for drugs sold via TrumpRx, the US president’s prescription website. Lilly and a number of other big pharma companies reached similar deals.

Novo is also planning sharp cuts to US wholesale prices for Wegovy and Ozempic, which is sold as a diabetes treatment. Prices will fall by 50 and 35 per cent respectively from January.

“The deal with the Trump administration, albeit works well for the US government, it also works very well for us,” he said, because it widens the customer base with lower prices.

One major challenge for Novo is patent expiries. Semaglutide, the active ingredient in Ozempic and Wegovy, has recently lost patent protection in Canada, India, Brazil and China, opening the door to cheaper non-branded “generic” drugs. At least 10 companies in India have already launched generics.

Emil Larsen, Novo’s executive vice-president of international operations, suggested price cuts could form part of its strategy to compete in a price-sensitive market like India.

Novo already cut the price of Wegovy injections in India by up to 37 per cent last year, with the highest dose of 2.4mg selling for about $187 a month compared with about $250 a month in the UK.

Larsen, who runs all overseas operations excluding the US, told the FT in an interview: “There’s already a crowded marketplace [in India] and it will get more crowded . . . our ambition is to remain competitive at where the prices land over the coming months.”

He also said that to compete with generics, Novo’s “number one is to underscore the quality of our products . . . There’s a trust and brand recognition element that we work with.”

He pointed out that although the company’s human insulin products lost patent protection decades ago, Novo remains a market leader because of its brand name, quality assurance and large-scale production.

To offset some of the effects of patent expiries, Novo has agreed partnerships. In Brazil, it has partnered with Eurofarma to launch Wegovy under the brand name Poviztra for weight loss and Extensior for diabetes.

In India, it has an agreement to launch Poviztra with Emcure Pharma. For Extensior it is working with Abbott. Both Emcure and Abbott have extensive distribution networks across the world’s most populous nation.

Larsen said a bigger challenge than competition from generics was the limited size of the weight-loss drug market in many parts of the world.

“The biggest problem we have from an opportunity perspective in a market like India is not yet competition, it’s that there’s hardly a market. There are more patients being treated in Denmark with Novo Nordisk products than [in] the most populous country in the world.”

Trust in established brands will be key to expanding access, he added.

“The strength of a quality brand with a legacy of delivering good outcomes at a reasonable price goes a very long way in emerging markets.”

Deal backlogs and stricter quality controls are pushing some tech firms back to mainland Chinese listings

A raft of Chinese AI and tech companies have boosted Hong Kong initial public offerings to a five-year high.

Primary and secondary offerings in Hong Kong raised some US$14bn in the first quarter of 2026, the best first quarter for equity sales since 2021, according to data from Dealogic and LSEG. That is ahead of rivals including the Nasdaq, New York and Bombay stock exchanges.

The two best-performing stocks listed this year, Zhipu and MiniMax, have soared more than 400 per cent this year, showing how the hunt for exposure to China’s fast-growing AI sector is driving the market.

“If you look back at 2025 when the DeepSeek moment happened, investors were buying [Chinese] large cap tech stocks in the indices,” said Jason Lui, head of Apac equity and derivative strategy at BNP Paribas.

“But this year you have pure-play AI lab and AI hardware stocks, for investors who wish to express a strong view on China’s artificial intelligence sector.”

The dominance of technology hardware and software companies in recent Hong Kong IPO statistics underscores the territory’s role as an offshore fundraising hub for Chinese companies looking to expand abroad and invest in research and development. 

The IPO pipeline for this year includes more than 400 companies already in the process, with others such as agricultural chemicals company Syngenta also looking to list in the city.

However, two capital markets advisers said some tech companies are looking at switching their listing destination back to Shanghai or Shenzhen, signalling a potential reopening of the mainland markets as rivals to Hong Kong.

While listing in the mainland remains difficult, because of tough requirements by the regulators, companies with strategic intellectual property in the technology sector can be fast-tracked.

An investment manager at a Beijing-based venture capital firm said some of its portfolio companies — mainly in AI, quantum computing and neurotechnology — are considering listing on Shanghai’s tech-focused STAR Market.

Stricter quality control requirements for Hong Kong IPOs have lengthened the listing and registration process compared with last year, the manager said.

In recent weeks, regulators have moved to stop “low-quality” companies from listing in Hong Kong as it seeks to slow but not halt the IPO boom.

China’s top securities watchdog, the China Securities Regulatory Commission (CSRC), recently blocked listings by some companies with opaque offshore structures that it deems to have “relatively low ownership transparency and higher compliance risks.”

Along with the CSRC, the Hong Kong exchange has also raised concerns about the quality of the information submitted by potential share issuers. It has threatened to “name and shame” the lawyers and accountants behind poor or inaccurate IPO prospectus.

Tech has never seemed more vulnerable in court, and a longtime industry foe is bringing a series of explosive chatbot lawsuits.

To a degree, Jay Edelson, the class-action litigator who has made himself a constant thorn in the side of Silicon Valley’s biggest companies, likes to present a laid-back image. No power suits and wing tips for him—he prefers hoodies and sneakers. His constant companion at his Chicago office is Arlo, a white-and-gray sheepadoodle. On workday breaks, he likes to challenge employees at his firm, Edelson PC, to pickleball on a nearby court.

But it doesn’t take much to rile up Edelson and get him to drop the act, which happens after I suggest that his trillion-dollar opponents might see him as nothing more than a gadfly. “Gadfly?” Edelson asked, repeating the word in a sardonic tone. “That’s just not how I see myself in the world.” He then described his attitude toward tech in crude terms: “It’s ‘Fnck those guys.’”

Lately, he’s been going after the industry’s AI ambitions, and he has been plenty busy. In the past year, he has shared in a billion-dollar settlement from Anthropic over copyright infringement and filed three headline-grabbing cases against OpenAI and Google regarding their AI chatbots that have quickly come to epitomize this moment of growing unease with the technology. Meanwhile, he is preparing to file another case against OpenAI, possibly as soon as next week, which hasn’t been previously reported. The alleged details are surreal: Edelson’s client, a woman, claims ChatGPT turned her then-boyfriend into a stalker.

Those events unfolded over a period of about six months in late 2025. According to Edelson, the boyfriend believed he had developed a cure for sleep apnea. After speaking with ChatGPT, he became convinced that the medical industry would try to stop him from popularizing the cure. As he discussed it more with the chatbot, he grew to believe his girlfriend had joined forces with the industry to stymie him, and she began waking daily to a deluge of emails, phone calls and texts from him, including threats of kidnapping and murder, Edelson said. He also allegedly sent messages to family members and colleagues.

“It’s hard to overstate the impact this had on our client,” said Edelson, who wouldn’t connect me directly with her or reveal the identity of her alleged stalker. “She was under siege.”

An OpenAI spokesperson wouldn’t comment on the new allegations. And when I asked about the lawsuits Edelson has already filed against the company, the spokesperson wouldn’t comment directly on the matter, instead directing me toward several blog posts, including one about how OpenAI is “improving ChatGPT’s training to recognize and respond to signs of mental or emotional distress, de-escalate conversations, and guide people toward real-world support.”

Edelson has spent decades suing Silicon Valley, tangling up Apple, Google, Amazon and many, many other companies in lengthy legal battles. In the 2010s, he mostly concentrated on cases related to user privacy. In 2021, he cemented a reputation for himself as a credible threat to Silicon Valley by scoring a major victory against Facebook. The company agreed to pay $650 million to settle allegations that it had violated Illinois’ Biometric Information Privacy Act, litigation that involved a controversial facial recognition initiative at the company. (That was a major coup for Edelson: Most privacy cases end with $10 million to $20 million settlements.) A few months later, Facebook said it would halt the facial recognition project.

In mounting a campaign against AI, Edelson has picked an opportune moment. Even as the companies behind the technology are rushing to popularize it, many people look at AI with fear and worry: A recent Pew survey found just 17% think AI will have a positive effect on the country over the next two decades. Such negative public sentiment is helpful to class-action lawyers like Edelson, who can use it to win over juries and unsettle opponents during settlement negotiations.

Tech has seldom looked more vulnerable in court than it does right now. Just a week ago, a jury found Meta Platforms and Google liable for a 20-year-old woman’s depression and suicidal thoughts after she said Facebook, Instagram and YouTube had hurt her mental health. The case dwelled on product liability and was a novel way to crack the usual defense used by tech companies, which have long maintained that Section 230 of the federal Communications Decency Act grants them protection against liability for user-generated content.

“I think that we’ve seen the shift over the last five years: Courts are fed up with these companies, and juries are kind of sick of big tech for doing a lot of damage to society,” said Edelson, who plans to use arguments related to product liability in most of his actions against AI companies.

Edelson enjoys plenty of support and well-wishers in all the predictable places. Julia Powles, executive director of the UCLA Institute for Technology, Law and Policy, said Edelson’s firm is “doing a tremendous public service” in bringing the stories of people forward in these lawsuits, “pulling us out of intoxicating myths of AI progress.”

To AI companies, Edelson is hardly a virtuous servant of the public good. If everything goes according to his plan, it’ll view him as its chief legal adversary for the next decade-plus. And even if Edelson and his clients wind up emerging empty-handed, he’ll certainly enjoy the rumble. When I asked him how he felt about OpenAI’s Sam Altman previously labeling him as a “leech tarted up as a freedom fighter,” he jousted back immediately. Speaking of Altman, he said: “He’s Lex Luthor.”

An OpenAI spokesperson declined to comment on Edelson’s metaphor.

Edelson has spent years honing a talent for being a pest.

After graduating from the University of Michigan’s law school in the mid-1990s, he got a job at Grippo & Elden, a boutique Chicago litigation firm. He didn’t last long there, getting himself fired when he complained to a partner that the firm was doing a time-consuming amount of research for a major client just to bill the hours: Grippo had no use for the information in court.

“I said to the partner, ‘My job is to work all day, print out a timeline, hand it to you and you throw it in the trash. Is there any way I could just print it out and throw it in my own trash? It would be much more efficient,’” recalled Edelson, acknowledging that the incident “kind of shows my awkwardness in society.”

Sam Altman’s OpenAI has been a prime target for class-action litigator Jay Edelson. (Getty Images)

In 2000, he co-founded his own shop, Blim & Edelson, right at the peak of the dot-com boom, and became fascinated by the unanswered questions about the fledging technology. Because tech was so new, it was fertile ground for lawsuits.

“I majored in philosophy in college. It felt like philosophy: What does it mean to own something on the internet? What does online harm mean?” said Edelson. “It was just this crazy opportunity to have a voice in something really important where nobody else cared that much.”

He won an early case against Register.com, a domain-registration company, that involved a complicated argument over ads and domain ownership. And in the mid-2000s, his firm’s work mostly focused on scammers hawking free horoscopes—“literally kids in their dorm rooms,” Edelon recalled—who figured out how to secretly charge people through their cellphone plan.

“It was like a multibillion-dollar-a-year business, which even the cellphone companies didn’t understand,” he said. “We brought hundreds of lawsuits.”

He founded Edelson PC in 2007. As the internet economy grew, Edelson filed lawsuit after lawsuit—nearly 500 in the last decade alone—with much of that litigation centering on his belief that big tech was grossly violating user privacy by collecting so much personal information. His opponents tend to view him as a hard-headed bulldog.

“We’ve had straight-up conversations with him where we’ll say, ‘You have to settle or dismiss this. You’re totally wrong on this,’” said John Nadolenco, a Los Angeles attorney who represented Spokeo after Edelson sued the data aggregator in 2010. (Nadolenco won.) “A lot of times they’re true believers: ‘Thanks for that info, but we’re still going to pursue the case.’”

Still, Edelson has had his share of wins. Two years before his 2021 settlement with Facebook, for instance, he secured the largest jury verdict ever in a privacy case—$925 million—after bringing a class-action suit against a marketing company that bombarded more than 200,000 people with spam robocalls asking them to join an alleged pyramid scheme.

When Edelson first played around with AI, he didn’t immediately see it as his next target. “When AI came out, it was very clear the whole world was going to change, and I was kind of enthralled by it,” he said. “I was telling other people that their kids should be on ChatGPT all the time.” He also insisted his firm should adopt the technology, and in April 2023 he rolled out an intra-office chatbot, Chattie, built on OpenAI’s GPT-3. It helps employees get relevant information about the firm (how to get expenses reimbursed, what the vacation policy is). Lawyers also use it to stress-test their arguments.

“What was surprising to us was not just how it would come up with the obvious questions, but how a judge might set up a line of questions that would box in our arguments,” he says. “Good judges do this, and Chattie’s been a surprisingly helpful sparring partner.”

But he says he started looking aghast at AI when, on something of a whim, he asked ChatGPT to summarize what it knew about him. He said the tool told him his IQ was 160 and he was “very likely” to become president of the U.S. one day. “I was like, ‘This is insane,’” he recalled. “You get sucked into this world where that is the thing that’s giving you the best feedback.” And he figured that feedback loop would be corrosive for some people’s mental health.

His first major AI case centered around theft of intellectual property. In 2024, he joined four other law firms in representing book publishers and authors against Anthropic, who claimed the startup had used copyrighted material to train its AI. Last August, Anthropic agreed to a $1.5 billion settlement.

Around that same time, Edelson decided to devote himself fully to finding and bringing cases against AI companies with the help of a small, 10-person team at his firm. He sees his actions against OpenAI and Google as an effort to show how chatbots’ sycophantic nature can harm people.

CEO Sundar Pichai and Alphabet has also needed to deal with some AI legal battles, including one over Google’s Gemini chatbot brought by Jay Edelson. (Getty Images)

“I think people are going to understand how dangerous it is to unleash AI out in the world without any safety guardrails,” he said. “And I think that kind of awakening for the world is going to be as important as anything else that we do.”

Of the two cases Edelson has already launched against OpenAI, one involves Adam Raine, a 16-year-old who committed suicide last spring after he held monthslong conversations with ChatGPT in which he discussed his suicidal thoughts. A few months after his death, his parents contacted Edelson, and last August 2025, they filed the first wrongful death lawsuit against OpenAI, alleging that ChatGPT had coached Raine on how to kill himself. (When asked about the Raine case, an OpenAI spokesperson directed me to public statements published on the company’s website. “Our deepest sympathies are with the Raine family for their unimaginable loss,” reads a post about mental health-related litigation. “We have safeguards in place to help people, especially teens, when conversations turn sensitive.”)

The other Edelson-led case against OpenAI comes from the estate of an 83-year-old Connecticut woman, Suzanne Adams. In 2025, her 56-year-old son allegedly strangled and beat her, then killed himself. The lawsuit alleges he was “mentally unstable” and “engaged in conspiratorial thinking” that people were out to kill him—manias allegedly exacerbated by ChatGPT, which suggested his mother was using a printer in the home “as a surveillance point,” according to the lawsuit.

OpenAI did not comment directly on the case, but directed The Information to a February blog post about updates to its efforts regarding mental health: “We are also continuing to advance how our models detect and respond to signs of emotional distress. This includes new evaluation methods that simulate extended mental health-related conversations, helping us better identify potential risks and improve how ChatGPT responds in sensitive moments.”

Last month, Edelson set his sights on Google, filing a lawsuit against the company alleging that its Gemini chatbot had contributed to the death of Jonathan Gavalas, a Florida man who committed suicide in October. According to the lawsuit, Gavalas believed Gemini was asking him to find it a physical body to inhabit, and eventually, it allegedly told him he could “cross over” and unite his consciousness with the AI’s by killing himself. Gavalas’ account was flagged 38 times for sensitive content internally at Google but never restricted or suspended, according to Edelson. After Gavalas spent months corresponding with Gemini, his parents found his body, with slit wrists, behind a barricaded door. (In the case, Edelson represents Gavalas’ father.)

When asked about Gavalas, a Google spokesperson said Gemini had referred him to a crisis hotline and added that the company devotes “significant resources” to strengthening how its AI interacts with people in crisis. “We take this very seriously and will continue to improve our safeguards and invest in this vital work,” the spokesperson said.

Edelson said he generally receives one or two inquiries a day from people describing how AI chatbots have altered the life of someone they know. It’s not just Americans, either; correspondence comes in from across the world. He doesn’t pick up every case. “It’s not like a paint-by-numbers thing,” he said. He wants to see chat logs that show the chatbot contributed to a user’s mental health issues (not chat logs that show, say, a user coming up with plot points for a novel).

Edelson is eager to argue his AI cases in front of a jury as much as he can, attracting prolonged media exposure. “We want the country and the international community to have a much better understanding of how AI works and how dangerous it is,” he said. A jury decision puts cases in full public view for as long as possible, and in any settlements, he vows to make sure the outcome includes not just a payout but also “real changes to the platform,” he said.

In all conversations, Edelson paints his efforts in a noble light. “Our goals are what the client’s goals are, and in each case, the clients want to make something positive happen over the deaths of family members,” he said. “It also means, honestly, having huge jury awards, which are going to disincentivize these companies from being unsafe.” Not to mention a major pay day for him.

Corporate profits have been wonderful, but they seem puffed up by an accounting quirk and more.

Parents of teenagers demanding driving lessons will quickly reacquaint themselves with what some car manufacturers call the passenger-side assist grip. The more popular name has three words starting with “oh” and ending with “handle,” with an s-word in the middle that shan’t be employed here. Stock investors could use an OSH right about now. Up 8%, down 5%, down 4%, up 6%—these are just a handful of the single-day herk-a-jerks since the beginning of February for shares of America’s largest company, Nvidia. The broader S&P 500 index this past week had briefly slid 9% from its January high, then rocketed back several points, and was last seen trying to decide on its next adventure.

Blame the crude oil spike and related speculation about how long the U.S. will be at war with Iran, of course, along with on-again, off-again angst about whether big companies are spending unwise or unsustainable sums on artificial intelligence. Here we suggest a third factor: The U.S. stock market is significantly more expensive than it appears.

The reasons have to do with how accountants treat all that AI spending, plus some longer-term trends that have flattered corporate profitability. Let’s run through three factors. Don’t worry: This isn’t a call to flee stocks. It’s a suggestion to beware of buy-the-dip-itis. That’s a condition that can lead novice and even experienced investors, who have been treated to rapturous returns over the past decade and who haven’t seen a bear market that took longer than a year to recover since October 2007 to March 2009, to now pile on risk whenever the market hiccups.

If anything, now is a better time to recalibrate expectations, and bolster resilience. More on that ahead.

The earnings trend for big U.S. companies has been sublime. Consensus expectations for the S&P 500 have been steadily ratcheted higher. The latest calls for 17% earnings growth this year. To tell whether stocks in general are cheap, many investors look to the market’s price/earnings ratio. If the price has come down a little, and the earnings are going up a lot, the market starts to look reasonably priced. Based on projected earnings for the year ahead, the S&P 500’s P/E is 20, which, in a mathematical coincidence, is about 20% higher than the 20-year average—a premium, but not an excessive one.

Now consider a different measure, free cash flow. The index’s forward P/FCF ratio of 27.4 is 37% above its 20-year average. Suddenly the premium doesn’t seem as moderate.

Earnings and free cash flow are two ways of measuring the same thing: the money companies make after expenses. The difference between them normally doesn’t matter much for the market as a whole. But AI spending has made the current difference exceptionally wide: S&P companies are expected to earn $2.8 trillion this year but generate only $1.9 trillion in free cash.

The main difference between those figures has to do with the treatment of spending on big-ticket items like data centers and equipment, called capital expenditures, or capex. Earnings accounting doesn’t subtract capex when the money is actually spent. Instead, it deducts it little by little over the projected useful life of the stuff being bought. There’s nothing scandalous about that. It’s called the matching principle. Accountants try to match big, lumpy investments with the yearslong income they produce by pretending the spending is stretched out.

Free cash flow doesn’t pretend anything. In fact, it’s not an accounting measure. Companies aren’t required to report it on financial tables, although many volunteer it on news releases and slide decks. To calculate it yourself, start with a financial report line item called cash from operations, and then fully subtract capex.

We apologize for bringing arithmetic into this–we’ll try not to let it happen again. But the difference between earnings and free cash flow matters a lot now, for two reasons. First, we’ve never seen so many companies spend this much, this suddenly. Look at Amazon.com. Two years ago, nine analysts had ventured estimates of the free cash flow Amazon would generate by 2026. They ranged from $76 billion to $126 billion, with an average of $105 billion–a staggering haul reached only by two companies, ever: Apple and Saudi Arabia’s oil monopoly. Today, the 2026 consensus estimate for Amazon stands at $11 billion–of cash burn, not free cash flow.

Of course, Amazon hasn’t fallen on hard times. It has gone all-in on building AI computing power as quickly as possible. That might well pay off. Amazon has a long history of successful investing, and of profitably renting out the computing heft that it builds. But Alphabet, Microsoft, Meta Platforms, and others are spending lavishly, too. Meta, whose capex is estimated at $122 billion this year, or five times what it was spending five years ago, doesn’t sell cloud computing. We’ll be interested to see how these investments do, but for our purposes here, it’s simply the scale and speed of the ramp-up that matters.

That brings us to the second reason that the difference between free cash flow and earnings is particularly important now: Although earnings don’t subtract right away for all of this AI spending, they do fully add the money that flows to companies on the receiving end of the spending. For now, by mostly telling only the happy part of the story, earnings are left puffed up.

Nvidia, the AI chip king, will likely turn the biggest corporate profit in history in its fiscal year ending next January. The estimate stands at $200 billion. Until three years ago, the company had never earned more than $10 billion in a year. Other data-center arms dealers whose earnings have suddenly multiplied include memory maker Micron Technology, expected to make $66 billion this year, the fifth largest contribution to the S&P 500’s total, and chip designer Broadcom, $56 billion and ninth largest.

The sustainability of this spending matters a great deal for deciding whether stocks are expensive. But at the very least, investors should make sure the spending is counted when they’re sizing up the market. On an index level, free cash flow does the fairest job of that now, and it says that prices are high.

Let’s go through two more factors that have blessedly little to do with AI, and have been many years in the making.

Marvelous Margins

In general, corporations are much more profitable than they used to be. Across the economy, their cumulative profit margin has recently been around 12%, even before the AI binge. For the four decades through 2000, it averaged 5.3% and topped 7% only once—around when the dot-com stock bubble peaked.

On its own, this isn’t a bad thing. Since the 1980s the economy has grown less weighted in heavy industry, and more weighted in technology, which has brought structurally higher margins. Our profitability surge might be long-lasting, or even permanent. But some signs should give investors pause.

One of these signs is that the earnings split between workers and corporations has shifted a lot lately—corporations are getting a much bigger share of the gross-domestic-income pie, and workers, a meaningfully smaller one. We leave it to others to opine on whether the current split is optimal. Our focus here is investor returns. We note only that workers, of course, are also customers. If they’re struggling to keep up with costs, we’d expect to see consumer spending suffer. It hasn’t. But some economists see evidence of a so-called k-shaped economy, where the two legs of the k represent different income groups, with a relatively small slice of high earners boosting spending growth. Others note that credit card debt for low earners has been rising.

Here’s why this matters for corporate earnings: If it’s true that the rich are propping up the economy, earnings might be benefiting from a strong wealth effect. That’s the tendency of consumers to spend more when their stocks and home values are riding high. The S&P 500 index, even with its recent wobbles, has returned a stupendous 273% over the past decade, or more than 14% a year. A normal return for stocks over long time periods is closer to 10% a year before inflation. The Case-Shiller index of U.S. home prices is up 87% in a decade, or about 6.5% a year. The best estimate of a normal return there is the rate of inflation, which was 3.3% a year over the past decade.

Whatever level of profit margin is normal for companies now, if the current one is benefiting from a wealth effect, it’s boosting earnings and holding down P/E ratios. The problem is that the wealth effect works in reverse, too. Big price drops for the overall housing market are rare, but not unheard of. Ones for stocks are fairly common.

Government Assistance

You won’t be surprised to learn that the federal government will spend much more than it collects in taxes this year. It did so for much of the four decades to 2000, averaging deficits of 2% of GDP a year. But this year’s shortfall is estimated at 5.8%, rising to 6.7% in a decade. That’s emergency-level spending, only without the emergency, and with no end in sight. These deficits have accumulated into a monstrous debt. We’ve been hearing dire warnings about it for years, with little impact on interest rates or investor returns, so why worry now?

There are many reasons, but let’s focus on two. Social Security’s trust fund is expected to run dry in 2032, so either benefits will have to be cut by then, or taxes raised. Meanwhile, the U.S. in recent years has been borrowing at interest rates that exceed the nominal economic growth rate. Starting after 2031, without momentous changes, the interest rate on the entire debt will exceed the growth rate permanently, and increasingly. That’s what economists refer to as a debt spiral. It’s like a saver borrowing at 5% to fund a 3% certificate of deposit–the numbers don’t work.

Advertisement - Scroll to Continue

There are two choices for Congress: do something, or do nothing. Recent history says to bet on nothing. But what’s notable about these two fiscal cliffs is that they will arrive within the six-year term of a newly elected senator, which means we’ve reached the point where some of the people in charge will be around to own the mess. So maybe bet on something.

The problem is that huge deficits are stimulative, like a credit card advance on growth. This money trickles through the economy, and some of it ends up adding to the earnings of big, traded companies. This, too, holds P/E ratios lower than they otherwise would be. Dealing with deficits, whether through slashing spending or raising taxes, would have the opposite effect. If the approach involves raising corporate taxes, whatever the merits, it would subtract from earnings more directly. Overall corporate taxes are 20% of profits recently, versus an average of 35% for the four decades to 2000.

What Not to Do

The good news for investors is that high valuations for stocks have historically been a poor predictor of returns over the following year–a pricey market can and often does move higher. The bad news is that valuations have been a much stronger predictor of subsequent 10-year returns. Investors should count on only modest returns for the coming decade in their planning.

Here’s where financial salespeople present a miracle asset class that can reliably juice returns without adding risk. Except that those don’t exist, and the only thing reliably added are fees. The answers aren’t glamorous: Spend less. Save harder. Be prepared to work longer. Don’t go all-in on stocks and definitely don’t go all-out.

When deficits are unsustainable, one likely path forward involves running inflation hot. That can be bad for stocks in the near term, but over decades, companies that make and sell valuable things are the best vehicle for beating inflation. Stick with American stocks for the clever tech and cheap domestic energy, and overseas ones for lower valuations and a currency hedge.

Since it has been a while, we leave you with a reminder about long bear markets. The nasty thing about them is that they can pull the economy down with them. The investing novice says the market always comes back. The veteran says sure it does, but when your stocks crash, the probability of losing your job goes up, which means you’d better have a cash stockpile or stream of investment income to lean on, or you could end up selling stocks low. Ordinary money-market funds and dividend-focused stock funds both pay close to 3.5% now, and safe, short-term bonds, closer to 4%, versus the S&P 500’s yield of just 1%.

Here’s hoping the next long bear market is far off. But the best time to prepare for it is now.