Autumn 2026 Global Markets Outlook - The Hurdle
Stocks at Records. Oil Above $100. The 10 Year Above 5%.
Something Has to Give...
Three months ago, in my Summer 2026 Outlook, markets had traded a fragile ceasefire as peace. Brent sat in the low 70s, gold had been punished, the S&P 500 was near 7,500, and a Federal Reserve free to threaten tightening had made discipline the word of the season.
I gave fragile peace a 55 percent probability and renewed conflict 25 percent.
The second scenario printed.
----------------------
Hormuz never normalized.
Brent returned above 100.
The Fed delivered its first hike since 2023.
The 10 year Treasury closed September 23 at 5.11 percent and the 30 year at 5.40.
Yet the Nasdaq made record highs this week, Bitcoin is roughly 41 percent above its July 1 close, and rising earnings expectations have compressed the S&P forward multiple even with the index near its highs.
▫️ Brent round tripped from roughly 72 to above 100.
▫️ The Fed raised its target range to 3.75 to 4.00 percent, unanimously.
▫️ The 10 year Treasury moved through 5 percent.
▫️ The S&P 500 still closed September 23 at 7,706.
▫️ Bitcoin went from roughly 60,000 on July 1 to 84,382.
▫️ AI linked investment grade debt trades around 115 basis points over Treasuries against roughly 78 for the broad market.
At 5 percent, the Treasury stops being background and becomes the benchmark every story has to beat.
The word of the summer was discipline.
The word of the autumn is the hurdle.
This is the quarterly mark to market, what the summer book got right, what its second scenario exposed, and how I would position when every story has to clear the same price of capital. 👇
🟦 1. The Scorecard, Marked to Market 🟦
An outlook that will not grade itself is marketing.
Here is my summer book at the September 23 close:
▫️ Fragile peace, persistent inflation as the 55 percent base case.
❌ Wrong on the geopolitics.
Hormuz never normalized and Brent climbed back above 100.
📝 Autumn action: promote escalation from tail risk to a peer scenario.
▫️ Conflict reescalation at 25 percent.
✅ Right, and underweighted.
The scenario printed, but the portfolio treated it like a runner up.
📝 Autumn action: keep geopolitical convexity without chasing the barrel after the move.
▫️ Tactical energy with a remnant retained for a fragile ceasefire.
✅ Right.
The remnant became one of the positions designed for exactly the quarter that arrived.
📝 Autumn action: trim strength above 100, keep insurance until physical Hormuz traffic normalizes sustainably.
▫️ Long duration underweight until growth cracks.
✅ Right for a third regime.
The 10 year at 5.11 and the 30 year at 5.40 turned the summer warning into the quarter's cleanest rates call.
📝 Autumn action: maintain. Reverse only on a real growth break, not because 5 percent looks high.
▫️ Dollar cash overweight and CAD underweight.
✅ Right after a volatile path.
The dollar is back near a two month high while the Canadian dollar sits near a seven week low despite expensive oil.
📝 Autumn action: keep dollar liquidity overweight and CAD underweight while the policy differential persists.
▫️ Bitcoin and gold held strategically through the cyclical sale.
✅ Right, with Bitcoin doing most of the work.
Bitcoin rose roughly 41 percent from July 1 through September 23.
📝 Autumn action: hold the strategic allocation, resize nothing upward into the rally.
▫️ Prefer AI adapters with provable margins over enablers carrying financing risk.
❌ Too early, therefore wrong for the quarter.
Semiconductors and infrastructure rallied while the financing stress appeared first in credit spreads.
📝 Autumn action: retire the blanket adapter preference. Split AI by cash generation, financing dependence and completed task economics.
▫️ Stablecoins as the defining structural story.
✅ Right on the infrastructure, incomplete on the competitive mechanism.
Binance invested 100 million dollars in Circle rather than simply replacing the issuer with its own dollar.
📝 Autumn action: distributors can disintermediate issuers, but they can also buy the economics they need.
👉 Net grade: the summer got rates, energy convexity, scarce assets and the dollar posture right, missed the geopolitical base case and mistimed the AI equity split.
The exits worked. The probabilities need more respect.
🟦 2. Macro, Welcome to the Hurdle Regime 🟦
The summer Discipline Regime was about a central bank free to choose inflation over weak growth.
Autumn is harder because growth did not provide the escape hatch.
August headline CPI held at 3.4 percent and core CPI slowed to 2.4.
Yet July PCE inflation was 3.7 percent, core PCE 3.3, and the September FOMC median still projects 2026 PCE inflation at 3.7 percent.
August payrolls rose 162,000 and unemployment stayed at 4.1 percent.
September business surveys then pointed to the fastest US private sector expansion since 2021.
👉 The autumn problem is not classic stagflation, it is an economy strong enough to keep earnings alive and inflation sticky enough to keep the hurdle high.
🇺🇸 The United States carries the strongest earnings engine and the highest discount rate, with the Fed projecting 2.3 percent growth and 4.1 percent unemployment for 2026.
🇪🇺 Europe is tightening into an energy shock. The ECB raised its deposit rate to 2.50 percent while projecting only 0.9 percent growth for 2026.
🇯🇵 Japan raised to 1.25 percent, its highest rate in 31 years, and the yen weakened anyway.
🇰🇷 Korea still owns one of the cleanest AI export stories while importing the energy shock that comes with it.
🇨🇳 China reaches the US summit with a powerful external engine, a multi year high yuan and still weak domestic demand.
🇨🇦 Canada held at 2.25 percent while the Fed hiked, leaving the currency on the wrong side of the differential.
➡️ The Hurdle Regime is selection made macro. Growth exists, but every asset now has to prove it can compound above a risk free alternative that finally pays.
🟦 3. The Warsh Fed, Guidance Meets the Long End 🟦
The summer said Warsh was raising the price of certainty. September added an actual rate hike.
On September 16 the FOMC voted 12 to 0 to raise the target range to 3.75 to 4.00 percent.
Sixteen of eighteen participants placed the appropriate year end rate at 4.125 percent or higher.
The projections explain why.
The Fed raised its 2026 growth median to 2.3 percent, lowered unemployment to 4.1 and kept PCE inflation at 3.7.
One part of my earlier Warsh thesis needs a correction.
The balance sheet review exists, but a rapid runoff regime does not yet. The September implementation note still maintains ample reserves, rolls over Treasury principal and allows purchases of short Treasuries when needed.
👉 The Warsh Paradox remains a strategic idea rather than the operating system.
Rates are the live tightening tool, while the long end is increasingly being set by inflation, issuance and demand for Treasury risk.
September 23 made that distinction visible.
Strong activity data and a poorly received five year auction pushed the 10 year to 5.11 and the 30 year to 5.40.
📝 The reversal trigger remains a material labor break with softer inflation expectations. One bad auction is a warning about market clearing, not a reason by itself to buy duration.
🟦 4. Equities, Earnings Are Clearing the Hurdle So Far 🟦
My summer neutral on broad US equities was too cautious.
The S&P 500 closed September 23 at 7,706 and the Nasdaq printed record highs earlier this week, despite a Fed hike and a 10 year yield around 5 percent.
The interesting part is valuation. Rising earnings estimates have pulled the S&P forward multiple back toward 19 times even with the index near its highs.
👉 The index is cheaper near a record because earnings rose faster than price. The problem is that Treasuries became more competitive at the same time.
At roughly 19 times forward earnings, the illustrative earnings yield is only a little above 5 percent.
That is not a breakeven rule, but it shows how little room exists for disappointment when the 10 year offers a similar nominal number without equity risk.
▫️ Quality with pricing power stays the core.
📝 Keep $MOAT and $QUAL, with the usual warning that $QUAL carries substantial mega cap technology exposure.
▫️ AI enablers remain investable, but no longer as one basket.
Muse, the semiconductor rally and Asian chip exports show demand becoming visible outside the training narrative.
📝 Keep profitable compute and power bottlenecks, separate them from projects that require permanently abundant financing.
▫️ Defense remains structural.
📝 Maintain $EUAD and $ITA.
▫️ Energy is tactical again.
📝 Trim strength, retain convexity. Sustained physical normalization through Hormuz is the exit.
🟦 5. Rates, Commodities and the Dollar, The Treasury Becomes the Competition 🟦
The summer thesis that long duration fails during inflation led selloffs has survived the oil shock, the Fed hike and the return of Treasury supply risk.
On September 23 the 2 year Treasury closed at 4.85 percent, the 10 year at 5.11 and the 30 year at 5.40.
👉 At 5 percent the Treasury stops being the risk free rate in the background and becomes the competitor every risky asset has to beat.
▫️ Treasuries, stay short until the growth trigger arrives.
📝 Keep maturities short.
Extend when employment deterioration becomes persistent, not simply because 5 percent looks historically high.
▫️ TIPS, short maturity still wins.
📝 Keep $STIP rather than buying long real duration into an inflation sensitive curve.
▫️ Credit, get paid for supply rather than fear it blindly.
AI linked investment grade debt trades around 115 basis points over Treasuries versus roughly 78 for the broad market, while projected hyperscaler issuance reaches roughly 420 billion dollars in 2027, around 60 percent above the 2026 estimate.
The spread is better read as the market charging for concentration and extraordinary supply than as proof of default risk.
📝 Prefer short investment grade carry through $VCSH and underwrite AI credit issue by issue.
▫️ Oil is a physical market with a diplomatic option embedded in it.
Brent is back above 100 while Hormuz traffic remains dramatically below pre conflict norms.
On September 21 only two commercial vessels were visible crossing, with the caveat that ships operating without transponders can escape the count.
📝 Keep the energy remnant. Durable physical reopening is the exit trigger.
▫️ Gold stays strategic.
📝 Hold $GLD. Add on evidence of falling real yields or policy credibility stress, not simply on another geopolitical headline.
▫️ The dollar still owns the cleaner rate differential.
The dollar is near a two month high while the Canadian dollar is near a seven week low.
📝 Keep dollar liquidity overweight and $FXC underweight until the Fed Canada differential narrows materially.
🟦 6. Digital Assets and the Permission Economy, Distribution Beat Legislation 🟦
Bitcoin rose from roughly 60,000 on July 1 to 84,382 on September 23 while the Fed hiked, the 10 year crossed 5 percent and the Senate failed to advance the CLARITY Act.
That does not make Bitcoin immune to rates. It does show that the summer's cyclical sale was not a structural verdict.
▫️ CLARITY failed, but the vote matters precisely.
The Senate voted 49 to 50 against advancing the bill, leaving it short of the 60 votes required for cloture. No comprehensive market structure framework was enacted, so the existing agency led regime remains the operating reality.
▫️ Stablecoins delivered the quarter's best correction to my June thesis.
I wrote that when the distributor decides to issue, the independent issuer risks becoming a middleman it no longer needs.
Then Binance put 100 million dollars into Circle.
👉 The distributor did not replace the issuer, it bought a piece of the economics. Disintermediation remains one path, vertical integration through ownership is another.
The GENIUS framework is expected to take effect on January 18, 2027 unless final implementing rules accelerate the statutory clock. Until then, distribution partnerships matter because the market structure is moving faster than the final rulebook.
📝 Hold $BTC as strategic scarce asset exposure rather than chase the quarterly move.
📝 Watch reserve income, partner rebates, settlement volume and customer ownership. The question is no longer issuer versus distributor, it is who captures the reserve economics and customer relationship after they negotiate.
The falsifier is operating rather than price based: if payment and settlement usage stall while partnerships keep multiplying, the infrastructure thesis has outrun adoption.
🟦 7. AI, The Cost of Capital Rose While the Cost of Intelligence Fell 🟦
This is where the autumn story becomes genuinely different from the summer one.
The financing bill arrived. AI related investment grade debt trades around 115 basis points over Treasuries against about 78 for the broad market, and hyperscaler issuance is projected around 420 billion dollars next year.
But the model layer moved in the opposite direction.
Anthropic released Fable 5.1 and Mythos 5.1 on September 1 as the same underlying model with different safeguard envelopes. Fable is generally available, Mythos remains restricted to vetted access. Anthropic estimates that lower cache pricing cuts typical Fable 5.1 billed workloads by 25 percent.
Then on September 22 it launched Opus 5.5, which Anthropic says performs at roughly Fable 5.1 level on most work while costing about 40 percent less to run than Opus 5.
👉 The price of capital is rising while the price of completed intelligence is falling. That is why the AI trade can survive a 5 percent Treasury longer than a simple capex bubble analogy suggests.
The releases also sharpen the Permission Layer. Fable 5.1 and Mythos 5.1 share the underlying model, while access and safeguards change the capability envelope. June already proved that access itself can be a policy variable when export controls briefly forced Anthropic to suspend Fable 5 and Mythos 5 before those controls were lifted.
For investors, three trades now sit inside one AI label:
1️⃣ Compute and power demand can grow if cheaper intelligence creates more completed work.
2️⃣ Model economics can improve even while nominal token prices fall, because retries, cache reads and intervention rates determine cost per completed job.
3️⃣ Leveraged infrastructure can still fail if utilization and contracted cash flow arrive slower than the debt.
📝 Keep $SMH, $CEG, $GEV and $CCJ as the value chain, but size them for wider financing spreads.
📝 Treat leveraged data center finance as a different asset from profitable semiconductor or power exposure.
If inference utilization and cash generation fail to accelerate while capex and debt issuance keep rising, credit is warning before equities admit it.
🟦 8. Scenarios, Because the Hurdle Can Fall Two Ways 🟦
The summer matrix caught the second scenario and weighted it too lightly.
Here is the autumn version:
1️⃣ Restrictive boom, expensive energy, the base case (45% Probability).
Growth stays positive, employment softens without breaking, Brent lives broadly between 90 and 110, the Fed keeps another hike live and the 10 year stays near 5 percent.
📝 Dollar liquidity, short TIPS, quality, defense, selected AI infrastructure and the energy remnant.
2️⃣ Energy détente, the benign disinflation path (30% Probability).
Hormuz normalizes materially, Brent falls below 85, headline inflation rolls over and the Fed can stop tightening without recession.
📝 Cut the energy hedge, extend duration gradually and let equity breadth widen.
3️⃣ Renewed supply shock (15% Probability).
Hormuz disruption worsens or Bab el Mandeb becomes a second material constraint, Brent moves through 120 and the long end reprices inflation before growth can absorb it.
📝 Energy convexity, strategic gold and short duration are the insurance.
4️⃣ Financing or labor break (10% Probability).
Employment contracts or Treasury and credit markets demand a materially higher clearing price, forcing deleveraging and slower investment.
📝 Raise cash first.
Extend duration only when the break becomes disinflationary rather than merely disorderly.
➡️ The probabilities are mine and they are for arguing with. Peace is no longer the default and escalation is no longer a tail. The portfolio has to survive both.
🟦 9. The Master Thesis and the Posture 🟦
Every section above is one story wearing different tickers.
The decade's infrastructure is being built under two constraints.
The first is the Permission Layer: a stablecoin can be frozen, a model can expose different capabilities to different users, regulation can widen or narrow market access, and a shipping lane can become a geopolitical permission system.
The second is the Hurdle: every asset, company and infrastructure project now competes with a US Treasury yielding around 5 percent.
A permissionless asset can still fall because cash pays. A frontier model can become cheaper while the data center financing it runs on becomes more expensive. A record equity index can coexist with wider credit spreads because equity prices today's earnings while bonds finance tomorrow's capex.
👉 The Permission Layer tells you who can switch the system off.
The Hurdle tells you whether the economics can keep it running.
That preserves the structural case for scarce assets without a central issuer level switch, gold and Bitcoin foremost, with the precision that matters: the property belongs most directly to self custodied coin and directly held metal, not automatically to an ETF share, exchange balance or custodial claim.
🔼 Overweight:
- Dollar liquidity and front end bills, with $UUP only as the currency expression.
- Short maturity TIPS: $STIP
Quality with pricing power: $MOAT & $QUAL
- Defense: $EUAD & $ITA
- AI shovels, power and grid: $SMH, $CEG, $GEV & $CCJ
▶️ Neutral:
- US broad equities near records: $SPX
- Bitcoin and gold at a strategic 5 to 10 percent combined allocation: $BTC & $GLD
- Short duration investment grade credit: $VCSH
- Tactical energy after the move above 100: $XLE
- AI related investment grade credit, selective issue by issue.
🔽 Underweight:
- Long duration sovereigns until growth actually cracks: $TLT
- Canadian dollar while the policy differential remains wide: $FXC
- Highly leveraged data center finance without visible utilization and contracted cash flow.
- Speculative growth and profitless technology that need falling rates to justify the valuation.
- High beta altcoins without their own operating demand.
----------
The lesson of the spring was velocity, how fast you recognize a regime.
The lesson of the summer was composure, holding a decade thesis through a quarter that punished it.
The lesson of the autumn is arithmetic. When the risk free alternative pays 5 percent, narrative has to become cash flow, utility or scarcity quickly enough to deserve the risk.
The summer taught the market who could survive discipline.
The autumn will show who can clear the hurdle.
NFA. DYOR.
@KobeissiLetter Time to stop promoting Anthropic costly business case and try supporting Nvidia and Jensen Huang to save the most strategic company of the AI race?..
Not a bad idea 👇
https://t.co/JPxVBEooqL
The Downfall Of Jensen And The Coronation Of Dario
How Huang Lost His Seat At Nvidia For Good And Anthropic Became a $1.8 Trillion Company In the Same Week
There is a strange timing in the recent history of 2 of the most powerful men of the AI era...
the same week the USA and China held what is probably the most important meeting between the 2 nations fighting for global AI and advanced Tech domination.
-----------------
🇨🇳 One wanted Washington to sell advanced chips to China.
🇺🇸 The other wanted tighter restrictions.
This week, Washington moved toward the second man's position.
And the second man's company had the biggest stock market debut in history.
In May, Jensen Huang was travelling along with Trump to meet Xi. By most accounts, Jensen was one of Donald's closest allies and his go-to voice on chips at the time.
Yesterday, Trump met Xi again. Huang was absent, a week after he was pushed out of Nvidia, the giant he built for 33 years, from a booth at a Denny's in 1993 to a $4 trillion company the whole AI boom runs on... to put it plainly: one of the most valuable and strategic companies in the world.
That same day, Dario Amodei's Anthropic began trading on Nasdaq and closed worth about $2 trillion. The seventh most valuable company in America, less than six years old, and still growing 1,200% a year.
In the same timeframe, Huang lost control of the company he built, keeping only his roughly $130 billion stake as a passive shareholder, while Amodei raised $81.3 billion in the largest IPO in history, ahead of SpaceX, in a market selling off almost everywhere else.
You can even add to the irony and the cruelty of the events for Huang: before any of this, Nvidia had agreed to buy $10 billion of Anthropic stock in the IPO. So Nvidia now owns a piece of the company whose CEO lobbied for the very restrictions now hurting Nvidia.
Of course, nobody has proved any link between these events... yet. But if you wrote this week as a movie script, nobody would believe it.
Here's how the week unfolded, and what it changes for $NVDA and $ANT. 👇
----------------------
Huang resigned as Nvidia's CEO on Nov 12. According to the Washington Post, his fall was set in motion by Marco Rubio, one of Washington's most influential China hard-liners, through a memo built on intelligence about Huang and Alice Huang, the Megaspeed executive US officials suspect of working for Beijing. The memo's findings leaked. Who leaked them has not been made public. China calls the spy claims a smear.
🔹The Bridge Was On The Plane🔹
Huang pushed for years to lift the bans on selling high-end chips to China. In late 2025 Trump allowed H200 sales to China, a big win for him.
In May, Huang joined the Beijing trip after an invitation from Trump.
On that same trip, a US official says, Rubio briefed Trump about Huang's link to the suspected spy. Trump denies he got that warning.
So the warning and the man it was about travelled to China together.
This time the White House listed Rubio, Lutnick, Bessent and Greer on the trip. Musk was expected too.
Huang was not. A US official said Nvidia "should not really have any bearing on the trip."
What a funny thing to say.
🔹The Hawks Got Their Rule Before Takeoff🔹
Monday, Nov 16, the day Trump left for Asia, Commerce added licence requirements for advanced chips shipped to Singapore, Malaysia, Thailand, Indonesia and Brazil, and for data centres anywhere that rent large amounts of compute to Chinese companies.
US-headquartered firms are exempt.
The rule took effect Thursday. The day of the Xi meeting.
Beijing said it was "strongly dissatisfied" and promised countermeasures.
Trump, the same Monday: "I don't think you want to insult China just before you get there."
Bessent had already blamed Lutnick's Commerce Department for being slow on Nvidia chip curbs.
There is no more one administration with one China policy but two camps facing each other, and the Post says Huang's exit gave momentum to the one that wants tougher restrictions.
🔹What Shenzhen Said About Chips: Nothing🔹
Xi met Trump at the airport, something he had never done for any leader since taking power. Then came an hour-long meeting, their third this year.
Chips did not come up in Trump's opening remarks or in China's readout... But the White House had not published its own readout by Friday morning.
The closest Trump got was on Fox News: Blackwell chips could go to China in two or three years, once they are no longer the most advanced.
🔹The Man Who Wanted Them Banned🔹
January, Davos: Dario Amodei said selling advanced chips to China was like "selling nuclear weapons to North Korea." A proper media and diplomatic bomb.
Huang once said he disagreed "with almost everything" Amodei says. That probably included this last quote.
Euronews described the Nov 16 rule as "the kind of restriction Amodei has long called for, and one Huang has spent years fighting."
Then the IPO.
Anthropic priced at $83, above its $70 to $78 range.
$81.3 billion raised. The largest IPO on record, ahead of Elon Musk's SpaceX in June.
It opened at $111.50 and closed at $97.17, worth about $1.8 trillion.
The seventh most valuable US company, less than six years old.
Ironicaly, Nvidia is one of the biggest buyers in the $26 billion private placement that runs alongside the IPO, about $10 billion, tied with Google. The deal was agreed before Huang fell.
Now, Anthropic has also committed tens of billions of dollars to buy Nvidia's most advanced chips.
If everything goes right for them, Anthropic is bound to be Nvidia biggest customer.
So every China restriction that hurts Nvidia's sales helps a company Nvidia now owns a piece of, and that company keeps buying Nvidia chips.
Nvidia lost its founder and its China bet in the same week. It kept a $10 billion seat at the winner's table... without the magic hand of Jensen to play the cards.
🔹Who Lost🔹
▫️ Nvidia: heading for a seventh straight weekly drop, its longest losing streak since July 2008, down 38% over that stretch. It beat earnings on Tuesday, and that did not stop it.
▫️ Huang himself: still Nvidia's largest individual shareholder, about 812 million shares worth roughly $130 billion. All of the money, none of the power.
▫️ Beijing: it lost its best-connected voice in Washington.
Craig Singleton of FDD calls it "Beijing's worst nightmare": China wants chip access it can negotiate with the president, and it may get restrictions "written into law by both parties." Democrats take both chambers of Congress in January.
▫️ Trump's personal channel: he said he "wouldn't mind" if Huang came back.
His own hawks just wrote the rule anyway.
🔹The Questions, Clearly Labelled As Questions🔹
▫️ Who leaked the memo's findings, and why that week?
▫️ Why did the chip rule land on the day Trump flew to Asia, and not after the summit?
▫️ Beijing's state media is already asking its own version. Global Times: "Is someone behind the scenes pushing an agenda and stirring up confrontation?"
Consider the source. Its answer is always Washington.
To be fair, being right about a policy for years is not the same as pulling the strings.
The timing is the story, not a hidden hand, but how do you join the winning side in less than a week?
🔹What I Am Watching (NFA)🔹
▫️ Holding Nvidia?
📝 The China upside used to run through one man's phone line to the president.
Price it without that line now. Good news, it was already off the books. Just be reasonable.
▫️ Holding or eyeing Anthropic?
📝 It won the week on politics and on its numbers. The politics can turn. Watch what it reports next quarter, not the first day's pop. Anthropic is still the same it was last week... for now.
▫️ Want one date?
📝 Trump and Xi are expected to meet again at the G20 in Miami in December.
If chips stay off the table there too, the hawks have won more than a week. If things start moving everywhere right before this new milestone, expect some new developments.
🟢 I would drop the "hawks won" read if the White House readout or Miami brings chip relief for China.
🔴 I would harden it if the new Congress writes the rule into law.
Huang spent years keeping a seat between Washington and Beijing.
He lost it the week Anthropic got one on the Nasdaq.
Nobody has to plan a week like this. Someone only has to pick the date.
Coincidence of the calendar, or did someone in Washington choose the week?
📚 Sources (Jensen Huang, the Trump-Xi summit and Anthropic's IPO):
1. Reuters, May 14, 2026: Huang joined Trump's Beijing trip after an invitation from Trump, Trump picked him up in Alaska; H200 sales cleared for about 10 Chinese firms, no deliveries
https://t.co/LZKIA5D3yh
2. NVIDIA, Nov 12, 2026: Jensen Huang resigns as CEO, Colette Kress interim CEO
https://t.co/eqhJ9Tr2tD
3. Washington Post, Nov 13, 2026: Rubio's intervention set Huang's downfall in motion and gave momentum to officials pushing tougher restrictions
https://t.co/iLdrNfq15H
4. CNN via KESQ, Nov 13, 2026: intercepted intelligence, Rubio's memo to a small group of senior officials, briefing aboard Air Force One in May
https://t.co/vbT1G6oXQI
5. Washington Post, Nov 12, 2026: timeline, Trump allows H200 sales to China, "a significant win for Jensen Huang"
https://t.co/IXLikMTUCf
6. MS NOW, Nov 16, 2026: White House delegation list (Rubio, Lutnick, Bessent, Greer), Rubio briefing in May and Trump's denial, Trump "wouldn't mind" Huang reinstated, Bessent vs Lutnick
https://t.co/m6fKMgAkmt
7. CNBC, Nov 13, 2026: US official, Nvidia "should not really have any bearing on the trip"; Musk joins
https://t.co/2IyBjHaIqt
8. CNA, Nov 20, 2026: chips absent from Trump's opening remarks and China's readout, Nov 16 rule and US-HQ exemption, effective Thursday, Beijing's countermeasures, "insult China" quote, Blackwell in two or three years, G20 Miami
https://t.co/BBfL2gqovK
9. China's Ministry of Foreign Affairs, Nov 19, 2026: official readout of the Xi-Trump meeting in Shenzhen
https://t.co/RYfnIjtSaD
10. The Straits Times, Nov 19, 2026: hour-long meeting, third this year
https://t.co/ENi3DVRSh8
11. CNA (Reuters), Nov 19, 2026: Xi's first airport welcome for any leader, Craig Singleton (FDD) on "Beijing's worst nightmare"
https://t.co/WH0nrHg8Ue
12. BBC, Nov 18, 2026: Huang advocated lifting chip bans, Democrats control Congress from January, Global Times question
https://t.co/ngacnFlXoI
13. Axios, Jan 20, 2026: Amodei, chip sales to China like "selling nuclear weapons to North Korea"
https://t.co/iP0JXWzLb3
14. Euronews, Nov 19, 2026: Huang "with almost everything", rule "Amodei has long called for", $83 vs $70 to $78, Nvidia's $10bn+ in the placement, longest losing streak since July 2008
https://t.co/smh4Epqbsd
15. CNBC, Nov 20, 2026: $81.3 billion raised, record IPO, open $111.50, close $97.17, +17%, about $1.8 trillion, seventh most valuable US company, 1,200% growth, Nvidia down 38% over seven weeks
https://t.co/1mFBV8OKAX
16. NBC News, Nov 16, 2026: Trump heads to China without Jensen Huang
https://t.co/YYcnVwMFSX
17. Yahoo Finance, Nov 2026: Huang keeps about 812 million Nvidia shares, worth about $132 billion, largest individual owner
https://t.co/1mNZR4AlUN
18. Fortune, Nov 12, 2026: Nvidia agreed to buy about $10 billion of Anthropic shares in the $26 billion private placement; Anthropic committed tens of billions to Nvidia chips; Nvidia a $4 trillion giant
https://t.co/sSsCTUMCLI
19. Axios, Nov 13, 2026: Nvidia's ~$10 billion in the placement ties it with Google as the largest investor in that deal
https://t.co/g5SQJgHjgE
The Downfall Of Jensen And The Coronation Of Dario
How Huang Lost His Seat At Nvidia For Good And Anthropic Became a $1.8 Trillion Company In the Same Week
There is a strange timing in the recent history of 2 of the most powerful men of the AI era...
the same week the USA and China held what is probably the most important meeting between the 2 nations fighting for global AI and advanced Tech domination.
-----------------
🇨🇳 One wanted Washington to sell advanced chips to China.
🇺🇸 The other wanted tighter restrictions.
This week, Washington moved toward the second man's position.
And the second man's company had the biggest stock market debut in history.
In May, Jensen Huang was travelling along with Trump to meet Xi. By most accounts, Jensen was one of Donald's closest allies and his go-to voice on chips at the time.
Yesterday, Trump met Xi again. Huang was absent, a week after he was pushed out of Nvidia, the giant he built for 33 years, from a booth at a Denny's in 1993 to a $4 trillion company the whole AI boom runs on... to put it plainly: one of the most valuable and strategic companies in the world.
That same day, Dario Amodei's Anthropic began trading on Nasdaq and closed worth about $2 trillion. The seventh most valuable company in America, less than six years old, and still growing 1,200% a year.
In the same timeframe, Huang lost control of the company he built, keeping only his roughly $130 billion stake as a passive shareholder, while Amodei raised $81.3 billion in the largest IPO in history, ahead of SpaceX, in a market selling off almost everywhere else.
You can even add to the irony and the cruelty of the events for Huang: before any of this, Nvidia had agreed to buy $10 billion of Anthropic stock in the IPO. So Nvidia now owns a piece of the company whose CEO lobbied for the very restrictions now hurting Nvidia.
Of course, nobody has proved any link between these events... yet. But if you wrote this week as a movie script, nobody would believe it.
Here's how the week unfolded, and what it changes for $NVDA and $ANT. 👇
----------------------
Huang resigned as Nvidia's CEO on Nov 12. According to the Washington Post, his fall was set in motion by Marco Rubio, one of Washington's most influential China hard-liners, through a memo built on intelligence about Huang and Alice Huang, the Megaspeed executive US officials suspect of working for Beijing. The memo's findings leaked. Who leaked them has not been made public. China calls the spy claims a smear.
🔹The Bridge Was On The Plane🔹
Huang pushed for years to lift the bans on selling high-end chips to China. In late 2025 Trump allowed H200 sales to China, a big win for him.
In May, Huang joined the Beijing trip after an invitation from Trump.
On that same trip, a US official says, Rubio briefed Trump about Huang's link to the suspected spy. Trump denies he got that warning.
So the warning and the man it was about travelled to China together.
This time the White House listed Rubio, Lutnick, Bessent and Greer on the trip. Musk was expected too.
Huang was not. A US official said Nvidia "should not really have any bearing on the trip."
What a funny thing to say.
🔹The Hawks Got Their Rule Before Takeoff🔹
Monday, Nov 16, the day Trump left for Asia, Commerce added licence requirements for advanced chips shipped to Singapore, Malaysia, Thailand, Indonesia and Brazil, and for data centres anywhere that rent large amounts of compute to Chinese companies.
US-headquartered firms are exempt.
The rule took effect Thursday. The day of the Xi meeting.
Beijing said it was "strongly dissatisfied" and promised countermeasures.
Trump, the same Monday: "I don't think you want to insult China just before you get there."
Bessent had already blamed Lutnick's Commerce Department for being slow on Nvidia chip curbs.
There is no more one administration with one China policy but two camps facing each other, and the Post says Huang's exit gave momentum to the one that wants tougher restrictions.
🔹What Shenzhen Said About Chips: Nothing🔹
Xi met Trump at the airport, something he had never done for any leader since taking power. Then came an hour-long meeting, their third this year.
Chips did not come up in Trump's opening remarks or in China's readout... But the White House had not published its own readout by Friday morning.
The closest Trump got was on Fox News: Blackwell chips could go to China in two or three years, once they are no longer the most advanced.
🔹The Man Who Wanted Them Banned🔹
January, Davos: Dario Amodei said selling advanced chips to China was like "selling nuclear weapons to North Korea." A proper media and diplomatic bomb.
Huang once said he disagreed "with almost everything" Amodei says. That probably included this last quote.
Euronews described the Nov 16 rule as "the kind of restriction Amodei has long called for, and one Huang has spent years fighting."
Then the IPO.
Anthropic priced at $83, above its $70 to $78 range.
$81.3 billion raised. The largest IPO on record, ahead of Elon Musk's SpaceX in June.
It opened at $111.50 and closed at $97.17, worth about $1.8 trillion.
The seventh most valuable US company, less than six years old.
Ironicaly, Nvidia is one of the biggest buyers in the $26 billion private placement that runs alongside the IPO, about $10 billion, tied with Google. The deal was agreed before Huang fell.
Now, Anthropic has also committed tens of billions of dollars to buy Nvidia's most advanced chips.
If everything goes right for them, Anthropic is bound to be Nvidia biggest customer.
So every China restriction that hurts Nvidia's sales helps a company Nvidia now owns a piece of, and that company keeps buying Nvidia chips.
Nvidia lost its founder and its China bet in the same week. It kept a $10 billion seat at the winner's table... without the magic hand of Jensen to play the cards.
🔹Who Lost🔹
▫️ Nvidia: heading for a seventh straight weekly drop, its longest losing streak since July 2008, down 38% over that stretch. It beat earnings on Tuesday, and that did not stop it.
▫️ Huang himself: still Nvidia's largest individual shareholder, about 812 million shares worth roughly $130 billion. All of the money, none of the power.
▫️ Beijing: it lost its best-connected voice in Washington.
Craig Singleton of FDD calls it "Beijing's worst nightmare": China wants chip access it can negotiate with the president, and it may get restrictions "written into law by both parties." Democrats take both chambers of Congress in January.
▫️ Trump's personal channel: he said he "wouldn't mind" if Huang came back.
His own hawks just wrote the rule anyway.
🔹The Questions, Clearly Labelled As Questions🔹
▫️ Who leaked the memo's findings, and why that week?
▫️ Why did the chip rule land on the day Trump flew to Asia, and not after the summit?
▫️ Beijing's state media is already asking its own version. Global Times: "Is someone behind the scenes pushing an agenda and stirring up confrontation?"
Consider the source. Its answer is always Washington.
To be fair, being right about a policy for years is not the same as pulling the strings.
The timing is the story, not a hidden hand, but how do you join the winning side in less than a week?
🔹What I Am Watching (NFA)🔹
▫️ Holding Nvidia?
📝 The China upside used to run through one man's phone line to the president.
Price it without that line now. Good news, it was already off the books. Just be reasonable.
▫️ Holding or eyeing Anthropic?
📝 It won the week on politics and on its numbers. The politics can turn. Watch what it reports next quarter, not the first day's pop. Anthropic is still the same it was last week... for now.
▫️ Want one date?
📝 Trump and Xi are expected to meet again at the G20 in Miami in December.
If chips stay off the table there too, the hawks have won more than a week. If things start moving everywhere right before this new milestone, expect some new developments.
🟢 I would drop the "hawks won" read if the White House readout or Miami brings chip relief for China.
🔴 I would harden it if the new Congress writes the rule into law.
Huang spent years keeping a seat between Washington and Beijing.
He lost it the week Anthropic got one on the Nasdaq.
Nobody has to plan a week like this. Someone only has to pick the date.
Coincidence of the calendar, or did someone in Washington choose the week?
@unusual_whales And Nvidia still put about $10B into Anthropic's IPO... Lost the founder who fought Dario, kept the stake in Dario.
Strangest hedge of the year?
@FeeBeacon Where's the 283K figure from?
Lookonchain had the full 5.2M leaving the deployer, but if only a slice hit the book that's even worse for holders: the pool died before most of it could sell.
Wrapped Monero on Hyperliquid went from about $494 to $1 in roughly 35 minutes.
Real Monero didn't move, but - here again - the hacker infected vault got hit...
What on-chain trackers saw:
5.2 million $XMR1 left the token's deployer wallet for a fresh address.
It was dumped on Hyperliquid for about $13.2 million USDC.
The proceeds became 7,467 ETH.
Four days earlier, money from the Lighter theft had bought about $3.5 million of $XMR1.
Fun fact: Proxonchain says the Kava exploiter parked funds there too, and that none of those wallets sold before the crash.
They chose it because nobody could freeze it... Nobody could protect it either.
🤷 Hyperliquid wasn't the weak point.
🤷 Monero wasn't either.
👉 The wrapper was: the tokens came straight from one wallet.
Nobody has explained yet whether that wallet's key was stolen or used by its owner.
A privacy coin you can only hold through someone else's keys isn't private money... It's a promise from whoever holds them.
If you hold wrapped coins, do you really know who controls the keys behind yours?
🚨HYPERLIQUID WRAPPED MONERO APPEARS TO BE GETTING HACKED!
@lookonchain flagged an attacker wallet that received 5.2 million $XMR1, dumped it on Hyperliquid, then used the proceeds to buy ethereum:native.
So far the address has spent $13.22M USDC for 7,467 ethereum:native.
$XMR1 is the Hyperliquid-wrapped version of Monero that lets people trade $XMR exposure on the exchange through the Wagyu bridge. It’s not official Monero.
📚 Sources (Wrapped Monero $XMR1 crash on Hyperliquid, and the hackers caught in it):
1. Lookonchain via @crypto_banter, Nov 20, 2026: attacker wallet received 5.2M XMR1, sold on Hyperliquid, $13.22M USDC spent for 7,467 ETH
https://t.co/bpqXngsYB9
2. ScamSniffer via @ExitTheIllusion, Nov 20, 2026: transfer from the XMR1 deployer wallet, about $13.3M USDC, XMR1 from about $494 to about $1 in 35 minutes, XMR near $475
https://t.co/vcEq5ONmaH
3. @sjbtc9, Nov 20, 2026: Lighter trader API key leak on Nov 16, about $10.39M moved out, about $3.46M spent on XMR1 on Hyperliquid
https://t.co/EAh8BtVVtX
4. @aml_crypto, Nov 20, 2026: about $3.5M from the Lighter attack converted through Wagyu
https://t.co/0gBJ6JitD9
5. @Proxonchain, Nov 20, 2026: Kava exploiter, Lighter drainer and Safe multisig drainer held XMR1, none sold before the crash
https://t.co/AJB3CF98eD
🚨HYPERLIQUID WRAPPED MONERO APPEARS TO BE GETTING HACKED!
@lookonchain flagged an attacker wallet that received 5.2 million $XMR1, dumped it on Hyperliquid, then used the proceeds to buy ethereum:native.
So far the address has spent $13.22M USDC for 7,467 ethereum:native.
$XMR1 is the Hyperliquid-wrapped version of Monero that lets people trade $XMR exposure on the exchange through the Wagyu bridge. It’s not official Monero.
📚 Sources (Sky's $2.5B USDC wallet behind DAI and USDS, North Korea and the freeze risk):
1. CoinDesk, Nov 19, 2026:
https://t.co/Gplf1fTgV0
2. Etherscan, Sky LitePSM USDC pocket:
https://t.co/8OI9I7QqQz
3. Sky chainlog: that address is MCD_LITE_PSM_USDC_A_POCKET
https://t.co/2NX8HxRj93
4. Cointelegraph, Nov 19, 2026:
https://t.co/pMCz4jagRW
5. US Department of Justice, Nov 12, 2026:
https://t.co/vxbFzfDCxZ
The Hacker-Infected $2.5 Billion Wallet Behind ethereum:0x6b175474e89094c44da98b954eedeac495271d0f That Circle Could Freeze
North Korea Used It To Dodge Stablecoin Freezes. Honest Holders Carry The Risk
$2.5 billion of USDC sits in one Ethereum wallet. It backs part of over $8 billion of DAI and USDS.
Wallets tied to hackers, phishing and sanctions pushed $518 million through it. $101.9 million links to North Korea.
The real issue? Circle can freeze an address, not a slice of it...
----------------------
The findings come from Inca Digital, via CoinDesk this morning. The target: Sky's swap window, the old MakerDAO module that trades USDC for DAI or USDS at exactly $1.
Circle can freeze USDC. Nobody can freeze DAI or USDS that way. That one difference turns the window into an exit.
No freeze has happened. At 13:30 Paris today, the wallet was not blacklisted.
So who actually carries this risk: the hackers, Sky, or the people holding DAI? 👇
🔹A Swap Window With No Doorman🔹
$USDC in, ethereum:0x6b175474e89094c44da98b954eedeac495271d0f out, one for one.
No ID, no screening.
Inca says 95% of the flagged swaps came through aggregators like 1inch and ParaSwap, not Sky's own websites, which screen wallets.
These are wallets linked to illicit actors, not proven stolen money. Inca says so itself.
🔹Bybit Showed The Route, Binance Showed Why🔹
In the 10 days after the $1.5 billion Bybit hack in February 2025, wallets funded by the loot swapped $91.5 million through the window.
Almost $30 million moved after the FBI had named North Korea.
Then Binance, Oct 29: $6.3 billion stolen, at least 55% of it stablecoins, per Abu Dhabi court papers.
Tether and Circle blocked more than $2.3 billion of it, per Binance.
Nothing says Binance money used Sky's window. But freezes work, so stolen dollars run toward whatever cannot be frozen... fast.
🔹One Address, One Switch🔹
The $2.5 billion sits in one wallet set up through Coinbase Prime. Coinbase pays rewards on the balance.
If a court asked Circle to blacklist it, everything stops: dirty dollars, clean dollars, swaps both ways.
Bluechip's stablecoin ratings director: "a low probability outcome, but it is not entirely hypothetical." His precedent: $12 million of USDC frozen in a Zama contract in May, on a California judge's order.
🔹Sky's Defense🔹
Flagged swaps are under 0.06% of the module's $930 billion plus lifetime volume, mostly from 2021 to mid-2024.
But governance can halt the module without the usual delay. Sanctions lawyers notice that kind of control.
🔹What I Would Check This Week (NFA)🔹
▫️ Hold DAI or USDS?
📝 Your 1:1 exit runs through that window. If it stops, you sell at market, not at a fixed $1.
▫️ Earning on sUSDS, or borrowing against it?
📝 Galaxy holds $100 million of it and takes it as collateral. Ask your platform what happens if the 1:1 swap closes.
▫️ Want to watch it yourself?
📝 Open the wallet on Etherscan and run USDC's isBlacklisted on it. One minute.
I would call this overblown if Sky screens the window and the next big hack skips it.
I would take it seriously the day a freeze request names that address.
Hackers flee the dollar that can be frozen. The dollars they leave behind wait in one wallet that can.
The risk is not the $101.9 million. It is the $2.5 billion next to it.
The stablecoin nobody can freeze is backed by a wallet anyone with a court order could.
If you hold DAI or USDS, would you rather Sky screens that window, or keeps it open and lives with the risk?
📚 Sources (Nvidia's $118 billion in guarantees):
1. NVIDIA, CFO Commentary Q3 FY2027, Nov 17, 2026:
https://t.co/q1iEMY7q28
2. NVIDIA, Q3 FY2027 results, Nov 17, 2026:
https://t.co/VPuReq9XQq
My Nov 17 post on the Fed and AI's off balance sheet promises:
https://t.co/zECw15Q3G9
The Fed Found $3 Trillion The AI Boom Keeps Off The Books
Whose Savings Carry The AI Bet, Hours Before Nvidia Reports
$680 billion borrowed for AI data centers this year. Five times all of 2023.
More than $3 trillion in leases, contracts and guarantees on top of that, at seven tech giants.
The Fed put both numbers in writing on Friday.
----------------------
The Fed's Financial Stability Report, out Nov 13, gave AI financing its own box. Its survey of 23 market contacts ranked AI as the top risk, cited by 87%, up from 50% in spring.
On Monday the Nasdaq fell 2.5% and now sits more than 10% below its Oct 6 high. The chip index lost 5.3%. Nvidia lost 4.1%.
This morning, Seoul's KOSPI closed below 5,000 for the first time since February, about 46% under its June peak.
Nvidia reports tonight. So where does the AI risk actually sit, and whose money is it? 👇
🔹What The $3 Trillion Actually Is🔹
Good news, for once we are not talking about debt... actually quite the opposite.
More than $1 trillion is data center leases signed but not started yet.
More than $2 trillion is take-or-pay contracts and residual value guarantees, including chipmakers guaranteeing what their GPUs will be worth as collateral.
The seven giants carrying those promises: Amazon, Alphabet, Broadcom, Meta, Microsoft, Nvidia and Oracle.
A giant signs a long lease, a smaller developer borrows against it to build. The giant's balance sheet barely moves.
Their actual debt should pass $800 billion by year end, per Fed governor Barr.
The promises already sit above $3 trillion, at face value.
🔹Who Holds The Paper🔹
▫️ The bonds: mostly mutual funds, pension funds and insurers.
▫️ The riskier slices of the vehicles building the sites: high-yield funds, life insurers, BDCs and private credit.
▫️ The largest US banks: close to $1 trillion of exposure to tech firms, developers, those vehicles and utilities.
AI borrowing is about 31% of top-rated US corporate debt issued this year. It was under 6% in 2023.
And the loop: over $100 billion invested by the giants in AI labs and cloud start-ups that spend it renting compute back from them.
🔹Who Cashes The Promises🔹
Those promises are someone else's revenue.
▫️ Landlords and developers collect the rents.
Blue Owl's funds own 80% of $META's Hyperion campus in Louisiana and will lease it back to Meta, with a residual value guarantee for the first 16 years.
Equinix, Digital Realty and Blackstone's QTS collect rent on long hyperscale leases.
▫️ Neoclouds collect on take-or-pay compute contracts.
$CRWV's customers, including Microsoft and Meta, buy reserved capacity this way, and Meta has committed up to $14.2 billion.
Nebius has a Microsoft contract worth about $17.4 billion, and a Meta deal worth up to $27 billion.
▫️ Power and equipment suppliers get paid to keep the sites running.
Constellation and Vistra have 20-year nuclear contracts with Meta.
Vertiv, Schneider Electric, Eaton and GE Vernova fill the power, grid and cooling orders.
Nvidia sells the GPUs, then sits on both sides: it also agreed to buy unsold CoreWeave capacity through April 2032.
The same names that cash these promises today take the hit if the demand behind them fades.
🔹What Warsh Said, Word For Word🔹
The Fed chair's bad case, on Monday: "expected returns could turn down and equity valuations could fall, leading to an economic slowdown and a more notable rise in unemployment."
A scenario, not a forecast. Markets still price a 95.7% chance of a December hike.
The money funding all this gets more expensive either way.
🔹Three Checks Before Tonight (NFA)🔹
▫️ Own an index fund?
📝 You hold the equity side.
Check your bond fund's top issuers too.
▫️ Chasing yield in high-yield or private credit funds?
📝 Look for data center project finance in the holdings.
▫️ Watching $NVDA tonight?
📝 Analysts expect about $109.1 billion in revenue. Read the guarantees as closely as revenue.
Not only the market is looking for cash, now they want insurance on top of that.
🟢 This stays a stock market problem if revenue keeps covering the leases and no guarantee gets called.
🔴 It becomes a savers' problem if the riskier slices lose value before the headline stocks do.
On Nov 6 I wrote that moving debt off the balance sheet does not remove the cash obligation.
On Friday the Fed counted it.
The selloffs priced the stocks: revenue, free cash flow, growth outlook. Numbers reported every quarter, with a price on them every day.
Nobody has priced the promises yet. They are not on the balance sheet, they don't trade, and they only turn into cash to pay if AI demand disappoints.
At that point they come due all at once, on top of the falling revenue.
Off balance sheet is an accounting term, not a hiding place.
Where do you think your own AI exposure sits: your stocks, your bond fund, or your pension?... Let's face it, if AI crashes, nobody is safe.
Nvidia now backs up to $118 billion of its customers' data centers.
It holds $53.5 billion in cash.
Last night's headline was revenue: $115.8 billion, up 103%.
One should probably pay attention to this commentary from the CFO: neoclouds and frontier labs are "growing faster than their balance sheets and long-term credit profiles can support."
So Nvidia steps in as credit support.
Up to $105 billion of the $118 billion is one site: SB Energy's campus in Ohio, built for Nvidia hardware and leased to OpenAI for 20 years.
Nothing is due yet.
The guarantees switch on building by building, the first in fiscal 2029, and Nvidia says it has paid nothing on them so far.
Yesterday I said to read the guarantees as closely as the revenue. Here they are.
The chip seller is now also the buyer's guarantor.
If a guarantee only bites when demand fades, should it count like debt when you value $NVDA?
Nvidia $NVDA says many AI clouds and frontier labs are "growing faster than their balance sheets" can support, so it is helping finance their buildouts.
Its maximum exposure on guarantees for partners rose to $118 billion from $108.5 billion last quarter.
$105 billion of that is credit support for SB Energy's 4.25-gigawatt Ohio campus that will host Nvidia infrastructure for OpenAI under 20-year leases.
Supply commitments also jumped to $390 billion from $279 billion, mostly multi-year memory purchases.
Nvidia says it has made no payments on the guarantees to date.
@XFenaux Premier trimestre sans Jensen aux commandes?
il est parti il y a 2 jours, je pense qu'il faut quand même pas lui enlever le crédit de ce trimestre 😅
Nvidia, $NVDA, Q3 Earnings Summary:
1. Record quarterly revenue of $115.8 billion, above expectations
2. Record data center revenue of $107.6 billion, above expectations
3. Record net income of $65.3 billion, above expectations
4. Record Q4 revenue guidance of $130 billion, above expectations
5. Announces new $150 billion stock buyback
This is objectively one of the most impressive quarters ever posted by a public company.
The AI Revolution is in full-swing.