US Session | Thursday 25 June 2026 The $400 Billion Rally Nobody Saw Coming Three Days Ago — Micron, Sandisk, PCE, and the Week That Changed Everything

Three days after the worst chip sector rout in three years, Micron delivered $41.46B revenue (+346% YoY), $25.11 EPS, and $50B Q4 guidance. Sandisk is up 15%. Gold is above $4,000 again. PCE printed 3.4% — marginally hot but broadly in-line. The Dow is at a record high. The week that looked like a market break is ending as a market validation.

Tuesday looked like a reckoning. The AI chip complex triggered circuit-breakers in South Korea. Sandisk fell 13.6%. Gold broke below $4,000 for the first time in 2026. Silver hit a seven-month low. The narrative forming — that the AI memory investment cycle had reached its mid-cycle plateau — was coherent and gaining traction. Then Micron reported.

Adjusted EPS of $25.11 against roughly $20.78 expected. Revenue of $41.46 billion, up 346% year-on-year, against consensus near $35.85 billion. Q4 guidance of $50 billion versus the $43 billion Street estimate. Gross margins approaching 85%. These are not merely good numbers. They are numbers that make the AI-capex-plateau thesis look, in retrospect, like what it was: positioning anxiety dressed up as fundamental analysis, triggered by a single SK Hynix production-scheduling decision and amplified by the worst possible market conditions for that narrative to land in.

Reuters put the size of the single-session chip rally at approximately $400 billion. Sandisk — Tuesday’s biggest loser at minus 13.6% — is Thursday’s biggest winner at plus 15%, now at $2,150 with Citi lifting its target to $2,500. Qualcomm added 6 to 14% on its own raised 2029 data-centre revenue target of $15 billion. The week that looked like a break is ending as a validation.

THE WEEK’S VERDICTS

AI CAPEX CYCLE  Not plateauing. Micron’s Q4 guidance of $50B and 85% gross margin is the definitive refutation. The Tuesday rout was positioning, not signal.

CORE PCE  3.4% YoY — marginally hot (consensus 3.3%), hottest since Oct 2023. But May is the peak of energy-driven inflation. Disinflation reasserts from June data forward.

USD  DXY softened slightly to ~101.5 post-PCE. Fed September hike odds eased to ~63% from 68%. Dollar’s extreme run pauses rather than reverses.

GOLD  $4,040 after Wednesday’s sub-$4,000 print. The geopolitical floor at $3,800-$3,900 held without being tested. Recovery confirms structural bid.

EQUITIES  Dow at record intraday high. S&P +0.4%. But Nasdaq -0.3% on Apple -4-5% (Mac price hikes). Non-tech leading is the real equity story this week.

CRYPTO  Bitcoin ~$61,000 — off multi-year lows. ETF outflows + AI sector rotation weigh. LTC near 52-week low at ~$40. Risk-on not yet transmitting to crypto.

PCE at 3.4%: The In-Line Miss That Didn’t Matter

May core PCE at 3.4% year-on-year was technically a tenth above the 3.3% consensus — the hottest print since October 2023 and the highest headline PCE (4.1%) since April 2023. On any other Thursday, a 3.4% core PCE from the Fed’s preferred gauge would be materially USD-bullish and risk-asset-negative. Today it landed as a non-event, which tells you a great deal about how Micron’s earnings reshaped the session’s context before the data hit.

The analysts who drove markets’ response to the PCE — and the dollar’s modest easing to DXY 101.5 rather than extension to 102 or beyond — did so on the basis of an expectation that has become consensus in the past 72 hours: May’s PCE represents the peak of the energy-driven inflation surge that the Iran conflict enabled, and the Strait of Hormuz normalisation that has taken crude from $107 back to $70.80 will produce measurable disinflation in June, July, and August data. That forward read is already baked into Fed expectations. The Q1 GDP revision to 2.1% (from 1.6%) was the offsetting positive. Durable goods orders falling 4.5% was a mild growth concern. The net: Fed September hike odds ease to 63%, the dollar softens modestly, and yields drift. Nothing is reversed. The macro regime is intact. It is just slightly less intense than it was at Wednesday’s peak.

Gold at $4,040: Why the Recovery Makes Analytical Sense

Gold reclaiming $4,000 and printing $4,040 on Thursday is not a mystery once you understand what took it below $4,000 in the first place. The $3,987 Wednesday intraday low was driven by three simultaneous forces: dollar at 13-month highs, risk-off from the chip rout, and pre-PCE positioning caution. When all three of those forces partially reverse — DXY softens slightly, chip sector rallies $400 billion, PCE in-line rather than sharply hawkish — gold’s recovery follows mechanically.

More analytically important: the structural bid held without being tested. CSFX had identified the $3,800 to $3,900 zone as the level at which central bank accumulation demand creates a floor independent of the Iran premium or the Fed’s dot plot. That floor was not reached. The $3,987 low was the market discovering where the floor is without going through it. The gold bull case — World Gold Council 1,231 tonne Q1 demand, central bank diversification away from dollar reserves — remains entirely intact at $4,040.

USD/CAD and USD/CHF: The Dollar Pause

USD/CAD at 1.4225 and USD/CHF at 0.8126 are both at or near multi-month highs that reflect the dollar’s sustained strength since Warsh’s hawkish hold — but both are slightly off Wednesday’s extreme peaks as the PCE in-line print and Micron’s risk-on impulse soften the dollar marginally. USD/CAD at 1.4225 is the Canadian dollar near its 14-month low, compressed between a hawkish Fed differential (Fed at 3.50 to 3.75% versus BoC at 2.25%) and oil below $71 removing the commodity support that structurally underpins CAD. The PCE in-line result was insufficient to trigger a sustained dollar reversal, so USD/CAD’s path of least resistance remains higher until either the BoC signals a hawkish surprise or WTI recovers toward $75 to $80.

USD/CHF at 0.8126 reflects the same dynamic with a safe-haven overlay now removed. The Iran ceasefire stripped the safe-haven CHF demand that had been a structural support for the franc since February. The SNB at 0.00% against the Fed at 3.50 to 3.75% is an enormous carry differential that was previously offset by CHF’s safe-haven bid. With that bid absent, the pair is tracking the pure carry dynamic. The 0.8190 to 0.8217 zone — last week’s 52-week high — remains the upside target unless the PCE’s marginal miss produces a more sustained dollar correction than the market has priced.

Sandisk at $2,150: The Week’s Full Swing

Sandisk’s recovery from Tuesday’s $1,985 low to Thursday’s $2,150 — a $165 per share recovery or roughly 8.3% from the low — traces the exact opposite path of what Tuesday’s AI-memory-plateau narrative would have produced if correct. On Tuesday, the market priced in a scenario where HBM4 demand was decelerating and the entire high-bandwidth memory complex was repricing lower. Citi’s target raise to $2,500 — which would represent a further 16.3% from Thursday’s level — reflects the opposite scenario: HBM demand is accelerating, Micron’s $50B Q4 guidance implies continued pricing power, and Sandisk’s own exposure to AI memory infrastructure positions it as a direct beneficiary rather than a warning signal.

The $2,100 to $2,200 consolidation zone is the technical range to watch for the rest of the session. A close above $2,200 confirms the recovery is complete and buyers are returning to previous highs. A close below $2,100 would suggest some residual anxiety in the sector despite Micron’s blowout, potentially from Apple’s separate negative signal — Mac price hikes suggesting margin pressure from AI component costs at the consumer end of the chain.

The Dow Record and the Apple Anomaly

The Dow Jones Industrial Average notching a record intraday high on the same day that the Nasdaq is down 0.3% is one of the week’s most informative signals. The Dow’s composition — heavy in financial, industrial, consumer, and healthcare names with less concentrated tech exposure than the Nasdaq — means it benefits from the same risk-on impulse driving chip stocks while being less dragged by Apple’s 4 to 5% decline. Apple is falling because its Mac price hike announcement — driven by AI component cost pressures — is being read as a margin warning from the world’s largest company by market capitalisation. That is an intra-tech divergence rather than a macro story.

The S&P 500 at 7,385 up 0.4% is splitting the difference. It has enough tech exposure to feel the Apple drag but enough broad-market exposure to benefit from the Dow’s tailwinds. The 7,300 to 7,350 support that was the risk-off concern through mid-week has not been tested. The 7,500 level — the pre-Tuesday peak — is the technical target for the recovery.

Bitcoin and Litecoin: The AI Rotation Problem

Bitcoin at approximately $61,000 — off multi-year lows but still under pressure — is experiencing the specific problem that occurs when a tech-sector rally is driven by an AI chips story rather than a macro risk-on story. The chip rally and the Nasdaq recovery are flowing primarily through the AI infrastructure chain: Micron, Sandisk, Qualcomm, Nvidia adjacents. Bitcoin is not in that chain. Crypto’s correlation with risk-on is strongest when the risk-on impulse is macro (lower yields, softer dollar, broad equity recovery) rather than sector-specific. Today’s risk-on is sector-specific. Spot Bitcoin ETF outflows are continuing, which means the institutional allocation trend that would provide a sustained floor for BTC at $60,000 to $62,000 is not yet asserting itself cleanly.

Litecoin near $40 — approaching the 52-week low at the bottom of the $40 to $44 demand zone — is the clearest expression of this dynamic. Fifteen consecutive sessions without spot ETF inflows. The structural demand zone is holding but without institutional participation. The Alpenglow and SPCX narratives that support Solana; the BitMine institutional accumulation floor under Ethereum — neither has a Litecoin equivalent. Patience here is the only defensible stance.

Read Full Report: capitalstreetfx.com/market-analysis/daily-market-analysis/