Here’s a preview of what we’ll cover this week: 

Macro: How’s The Economy Doing?; The Fed Gets Some Help

Markets: 13F Insights; Blood in the Streets; The Sellers Are Running Out; What Happens After a Friday Like This; The New Deal; Nvidia: Can’t Get More Bearish; HubSpot at 12x: The AI Casualty That Isn't

Lumida Curations: Nir Bar Dea Explains Bridgewater’s Research Culture; Jason Calacanis On AI in Netflix's New Fincher Film

Spotlight

On Thursday, I sat down live with Kartik Chopra, former CIA officer and now Co-founder and CEO of Adelphi, which sells AI datalakes for intelligence agencies.

Kartik was recruited to the CIA straight out of college.

Here's what we covered:

  • What export controls mean for Nvidia and the AI trade

  • Where China is ahead of the US in AI, and why

  • The state of American defense tech

  • Aliens (All I could get was a ‘yes’. :)

People say there's no edge in information anymore. Spend an hour with a former CIA officer and see if you still believe that.

The feedback has been great.

The full conversation premieres on our YouTube channel this Monday, November 23. Watch it here and set a reminder so you catch it when it goes live. 

You can also find it on Spotify.

For more conversations like this one, subscribe to our YouTube channel.

Hedge Fund Money Moves In 3Q26

Lumida tracks some of the best and brightest hedge funds, including specialist funds that most people have never heard of.

Our 13F grid records the top five positions, biggest additions and biggest reductions for 50 funds along with our read on where the money moved. 

It also includes every quarter back to 2023, so you can see how each manager's thinking has evolved.

You can download our Q3 2026 13F analysis grid here. 

You can track hedge fund positioning in the Lumida Invest app as well. The app tracks what each fund is buying, selling and holding, with an AI summary on the broad positioning. Check our app today.

Here are some insights from the recent 13F filings. 

Everyone Was All In

In 2Q26 highlights, we saw there was really only one trade in hedge fund land: AI.

The Q3 filings show how far it went. 

On September 30, 34 of the 50 funds that we track held Nvidia or one of the four hyperscalers among their top five positions, up from 26 a quarter earlier. 

Even the contrarians like David Tepper (Appaloosa) were in.

When everyone is already long, there's no one left to buy. 

And, it doesn't take much bad news to tip a crowded trade over. 

And, the last two months delivered plenty: sticky inflation, a rate scare, the midterms, concerns around capex growth, and then Jensen’s personal issues.

One of the most-bought stocks of the quarter was Intel. Fifteen funds bought it and only two sold. 

Soros made it his largest position, and Lone Pine and GQG Partners opened new stakes.

If you'd told a hedge fund manager two years ago that Intel would be the elite's favorite stock, he'd have asked whether you were also long Blockbuster.

But, the buy on Intel hasn’t worked. The stock has since more than halved, including its worst day since 1974.

Meta tells a similar story. 

This quarter, it became a top-five holding at 11 funds, up from 4.

When a name goes from 4 top-five spots to 11, you're no longer early. You're the crowd. 

Meta is down about 23% since September 30, and the debate is the same one we flagged in October: capex is running ahead of free cash flow.

Different Seats Within The AI Roller Coaster

Inside the chip trade, managers started picking which part of the stack to own.

Renaissance sold 99.7% of its Nvidia, which had been its largest holding, and grew its Micron position about sevenfold.

Bridgewater and Soros made the same switch on a smaller scale.

Whale Rock and Atreides went the other way. 

Whale Rock, a tech specialist we also like to examine, cut SanDisk, its largest position, by about two-thirds and quadrupled its Nvidia stake. 

Atreides cut Micron by more than half and added Nvidia.

Both camps had good arguments. But, in the end, it didn't matter.

Since September 30, Nvidia is down 45% and Micron 44%. The swappers changed seats on the same roller coaster.

The only trade that really worked was getting off the ride. 

Tepper vs. Leopold

This brings us to the two most-watched AI bulls in the filings.

At the end of May, Tepper's (Appaloosa) book had started to look a lot like Leopold Aschenbrenner's. 

Tepper owned Micron, CoreWeave, Taiwan Semiconductor, and Vistra Corp, much like Situational Awareness. 

However, in this quarter, they had split.

Leopold sold every share of Micron and SanDisk, $11.2 billion, or 56% of his book. 

He named his fund Situational Awareness, and apparently the situation called for an exit. 

He did keep call options on SanDisk and the DRAM ETF. 

But, Tepper walked further in. 

Micron was Appaloosa's largest position at 17.6% of the book in 3Q26. 

There's no China and no energy in the portfolio, and he rotated out of AMD into Nvidia.

I would not have guessed Appaloosa would go all in on AI.

If you squint, you could confuse it with Leopold's portfolio from six months ago.

But remember who Tepper is. 

He made his name in 2009 buying the banks that everyone at the dinner party knew were insolvent.

We're in Tepper's seat on this one. We own AI names, like Nvidia, TSM, and Micron and believe they offer compelling value if you can hold your nose and wait one year. 

Deleveraging

In our 13F review last November, Google was the top pick across hedge fund land.

This quarter, it was the stock everyone sold.

All of this is consistent with deleveraging. Hedge fund long/short exposure is now at multi-year lows. Tech valuations are also at multi-year lows. 

Alphabet had 14 sellers of the 50 we track, the most of any stock, and it dropped out of the top five positions in eight funds. 

We believe Google is now a strong buy, and we added recently.  

At 19x forward earnings and 20%+ earnings growth, Google is an interesting opportunity, and we are overweight. 

Druckenmiller's Next Moves

Last November, Druck opened Amazon, Meta and Alphabet.

This quarter, he took the opposite turn.

He sold 94% of his Taiwan Semiconductor trade and built Seagate up to 7.6% of the book, plus calls.

So he was still long the AI trade, but with a different horse.

He also opened a basket of gold miners (Newmont, Barrick and Agnico Eagle) and exited Alphabet, SanDisk, United and Delta.

He also added AppLovin. Druck is experiencing one of his largest corrections out there.

Net-net, Druck’s 13F says he wants exposure to:

  • AI storage rather than AI manufacturing

  • Gold, as a hedge against inflation

  • A few concentrated, high-conviction growth names (Natera, AppLovin)

We agree with him on AppLovin and storage.

Elsewhere in Hedge Fund Land

Warren Buffett trimmed Bank of America, doubled Lennar and opened Bending Spoons. 

Not owning AI turned out to be the best AI trade of the quarter.

That said, we see Apple sporting nosebleed valuations and don’t believe it has much upside from here. It’s been a defensive crowded name - like Costco or Walmart.

When the AI names recover, these should give back. 

Steve Cohen (Point72) picked up Celestica and Medtronic. We own both.

Blood in the Streets

The Nasdaq 100 fell more than 10% this week, and the S&P 500 declined 8.6%.

It was the S&P's worst week since April 2025, and the Nasdaq is now negative for the year.

The last time we saw the VIX hit 40 was also in April.

That day also marked the bottom in stocks.

What added to the volatility this week was (i) VIX expiration on Wednesday, and (ii) options expiration on Friday.

You can see the panic in the VIX. It touched 40 on Friday. 

The last time it got there was April 2025, at the tariff panic lows, and that turned out to be the bottom before a very strong rally.

We believe that will be the case here as well.

What's striking is that this selling isn't about earnings. 

Nvidia raised its growth guidance from 70% to 75% — despite concerns growth would lower due to China export controls.

We haven't seen missed estimates yet. 

The bear case now has to argue that earnings will slow. There’s no evidence of that yet, and the Texas Governor and others have indicated 2027 is in sight.

I felt the mood shift in person on Thursday night. 

At our event with Kartik Chopra, the room included a few professional money managers who had been constructive on AI for years. 

They had thrown in the towel this week, and not just on vibes. They had clear, logical arguments for why it was over.

That took me back to a party I attended in March 2009. 

Everyone in the room knew the banking system was insolvent. 

Equity investors had become overnight experts in CDO-squareds and counterparty risk.

The arguments were airtight. 

I remember thinking: everyone here has the same view, so it's time to go to the other side.

The market bottomed that month.

Today, everyone at the party is an expert on export controls and Singapore revenue and Jensen’s memes are aplenty.

The Sellers Are Running Out

The positioning data tells you how far this sentiment has gone, and how widespread it is.

Hedge funds have sold tech stocks for nine straight weeks.

The last six weeks of selling in tech is the fastest in a decade, led by semis and chip equipment.

Net leverage among fundamental long/short funds has fallen to 41.6%. 

That's below the Liberation Day lows and the lowest in more than five years.

The machines are out too.

CTAs sold more than $22 billion of S&P futures in a single week and are now net short, their lowest positioning since April. 

You can see the same sentiment in fund managers. 

Fund managers slashed their tech overweight from 37% to 12% in one month, the biggest cut since 2006. 

Retail sentiment looks the same, with bull/bear ratios in both the AAII and Investors Intelligence surveys now below their long-run averages.

Investors who levered up on the AI trade are getting margin calls and deleveraging.

The retail investors in 3x SOXL type semiconductor and memory funds are part of the deleveraging. When the deleveraging clears, like a forest fire that burns away the dead and fallen wood, the stage is set for growth again.

Valuations have already adjusted. 

The S&P 500's forward P/E has dropped to 17.6, below last year's Liberation Day low. 

These are excellent valuations and should generate attractive returns.

Profitability, meanwhile, is near record highs. Today's valuations already price in a decline in profitability that hasn't shown up.

Two caveats keep us from calling the exact low. 

First, fund managers sold tech but didn't raise cash, which actually fell to 3.6%. That’s the bear case.

Second, Goldman estimates about 75% of short-term trend signals are still negative, so systematic funds could keep selling in the near-term.

What Happens After a Friday Like This

Friday's 3.7% drop took the S&P 500 to its lowest close in three months.

We looked at every Friday since 1990 when the S&P fell 2% or more to a three-month closing low. 

There have been 19 such cases before this one.

The next session is usually rough. 

On average, the S&P fell another 1.3% the following day and finished higher only 8 times out of 19. Monday could bring one more flush.

Two months out, the picture flips. 

The S&P was higher in 16 of the 19 cases, with an average gain of about 8%, compared with 1.6% in a typical two-month stretch.

The last time this setup triggered was April 4, 2025, two days into the tariff panic.

We came across another study from X/OddStats below.

In the past 12 trading days, the S&P 500 has had five 1% drops, three 2% drops, two 3% drops and a 4% drop. 

That kind of cluster has only shown up during the market's scariest moments: Black Monday, the financial crisis, the 2011 downgrade, COVID and Liberation Day.

But, the AI moment is about deleveraging and concerns around Blue State policies.

(If you squeeze in one state, the datacenters move to another state - or country… That’s why we saw Trump pushing back on datacenters going offshore.)

Owning stocks on those dates felt terrible. A year later, the S&P was higher in 8 of the 10 cases, with a median gain of about 26%.

The results of these studies also line up with everything else we're seeing: panic in volatility, historic selling in tech, and positioning at multi-year lows.

There's a policy angle too. 

Policymakers respond to voters, and voters have 401(k)s.

Washington is watching this selloff as closely as you are.

How should you play it? Don’t panic, and don't try to nail the exact low. 

Use the fear to build positions in companies with good fundamentals, strong earnings growth and reasonable valuations. Those are on sale across many sectors right now.

When it feels easy to invest, it usually isn't a great time. 

When you feel a pit in your stomach, it usually is.

As Baron Rothschild supposedly put it after Waterloo: “Buy when there's blood in the streets.”

Then go for a walk, get some sun, and maybe stay out of your brokerage app until after Thanksgiving.

Macro

How’s The Economy Doing?

The economy hasn’t missed a beat.

Third-quarter GDP is now growing at 3.2% YoY. 

Business investment is doing the heavy lifting at about 9%, with equipment spending up 12%. 

Consumers are spending too, up 2.5% YoY.

Look at the chart below.

Consumers are strong because the checks are flowing.

The job market is holding up. 

Initial jobless claims came in at 203,000, near multi-year lows. Job openings are near highs.

Continuing claims also dropped to 1.76 million, significantly below their 2025 levels. 

We don't see a recession in the data. 

It's a services-driven economy, powered by baby boomers who have Social Security, Medicare and record household balance sheets behind them. 

Their spending doesn't stop because the Nasdaq had a bad month.

The Fed Gets Some Help

A strong economy has a downside: it keeps the Fed in hiking mode.

The October FOMC minutes show the committee held rates only to buy time. It wanted to assess the AI investment cycle, the effects of September's hike, and geopolitics. 

Since then, growth has stayed strong and inflation pressures have built.

The hawks are getting louder. 

Christopher Waller (Governor, Federal Reserve) wants to reverse all three of last year's rate cuts and says the Fed "may need to do more." He floated a 50bps hike in December. 

But, rate hike odds changed on Friday.

The chance of a 50bps hike has fallen from about 12% a week ago to zero, and we see some odds given to a hold from none last week.

Two things are taking pressure off.

First, the AI buildout is slowing. 

With Nvidia no longer backstopping financing, data center spending comes in at a lower slope. 

Less AI capex means less strain on chips, power and construction labor, and somewhat less inflation.

Second, oil is coming down. 

Brent has dropped from about $118 at the end of October to around $92, its third straight weekly decline, as Middle East exports recover and inventories build.

That flows straight into gasoline prices and inflation expectations.

The Fed still hikes in December. 

But slower AI spending and cheaper oil are doing some of its work for it, and that takes the 50bps scare off the table.

Markets

The New Deal: How Blue States Will Embrace AI

Utility bills are going up. 

The bad guys are the datacenters. 

Blue-state governors are going to cut a new deal with datacenters. 

The new “New Deal”:

1) no sales tax breaks and no incentives…

2) generate revenue for the county. 

3) add energy to the grid

Here are a few deal templates:

1) Loudoun County: $1.3 billion in tax revenue for a negligible portion of land. 

Loudoun County isn’t the only example. 

2) In Michigan, Google is adding to the grid. 

Google agreed to a 20-year power contract and agreed to fund up to 1.6 GW of renewable generation. 

3) In New York, Gov. Hochul said in July ‘I’m saying bring your own power, contribute more, give more to the community and you shouldn’t be getting tax breaks.’

The tax rebates will go away. 

The datacenters will be built. 

Governors will frame it as a win for consumers. 

4) Texas is one of the first states to enact legislation along these lines in January…

Texas Governor-elect Gina Hinojosa pledged to pause approvals for new data centers when she takes office January 19. 

For the January session, she wants lawmakers to:

- end data center tax subsidies

- shield households from electricity cost increases

- and give local officials the power to negotiate with developers or reject a project. 

The datacenter companies did a great job location scouting for incentives. 

The new deal will give the datacenters, consumers, and politicians a win. 

Policy uncertainty will reduce. 

The chips will flow.

Nvidia: Can’t Get More Bearish

Nvidia fell 14% on Friday to a new 52-week low. 

It was the stock's seventh straight weekly decline, its longest losing streak since 2008.

It's now down nearly 50% from its October high.

The trigger was a Reuters report that SEC enforcement staff are examining potential accounting issues at Nvidia. 

The same week, California's attorney general said his office is investigating whether Nvidia or Jensen's actions let China obtain advanced chips, and whether investors were misled about that risk. 

That came on top of Treasury Secretary Bessent's comments last Sunday, when he said Nvidia appears to have a culture of non-compliance.

These are serious words, and the regulators are doing what they must do. 

That's how the law works.

It's worth keeping three things in mind. 

First, an investigation is not evidence of wrongdoing, and many close without any enforcement action. 

Second, these take time, much longer than three months.

Third, Trump controls the executive branch - and that includes enforcement actions. He also is calling for Jensen to return as CEO. (We doubt that happens with these overhangs).

The board has hired an outside law firm, and the findings will fall into one of three buckets.

The worst case is a willful violation. That’s serious. It’s also priced in. 

In the middle is negligence: weak oversight and sloppy order forms, like the stadium security guard in the meme who waves everyone through. 

The best case for Jensen is that Nvidia had real controls in place and was deceived by Chinese buyers.

Historically, when an investigation like this wraps up, the overhang lifts and the stock moves higher in a short time frame. 

The quarter was excellent.

This was also the week when Nvidia reported a blockbuster quarter, but investors seemed to have missed it completely. 

Revenue came in at $115.8 billion, +103% YoY, well ahead of the $109 billion consensus. 

Data center growth accelerated to 21% quarter over quarter, and Nvidia guided next quarter to $130 billion, against $125 billion expected.

The number that mattered most was next year's outlook. 

Going into the call, the biggest risk was that Nvidia would cut guidance after shutting down questionable sales channels in Singapore and Malaysia. 

Instead, it raised the growth forecast from 70% to 75%. 

The second piece of good news was capital return. 

Nvidia will now send all excess cash flow back to shareholders, net of strategic investments, with about $227 billion left on its buyback authorization. 

Our one complaint is the pace.

Management plans to spread the buybacks out through the end of next year. 

Why buy back stock in an orderly fashion when it's this cheap? 

Concentrate the buying now, not when the stock is back at 18x earnings.

At this valuation, buybacks are the highest-return use of Nvidia's cash. 

Another area of focus on the call was Nvidia’s backstopping for the neoclouds. 

The takeaway was: the Bank of Nvidia is tightening its lending standards.

Nvidia's commitments to customers and suppliers jumped to $518 billion from $366 billion last quarter. 

But interim CEO Colette Kress says its guarantees only kick in once a project has land, power, a building and customers lined up. 

Basically, Nvidia now wants to see your pay stubs.

Colette Kress also said this isn't new, just clearer. 

Either way, it's the sensible move for an interim leader: don't change the strategy, reassure customers, shareholders and employees.

Nvidia’s Forward PE ratio is 8.7x.

It’s trading like a boring dinosaur company.

That’s the lowest valuation in at least a decade. Our read here, again, is forced deleveraging. 

For most of that stretch, it traded between 25x and 55x. The S&P 500 sits at 17.6x.

So the market is paying half the index multiple for a company that just raised its growth outlook to 75%.

The market is pricing a collapse in the business, and the guidance doesn't show one.

Sure - capex should slow down on the margin. But, not by 50%.

At 8.7x, you're being paid to wait.

HubSpot at 12x: The AI Casualty That Isn't

The market has decided AI will eat HubSpot.

The stock has fallen about 60% over the past year, and the story behind that drop is simple: AI agents will replace the CRM, and software seats will disappear.

The third quarter tells a different story. 

Revenue grew 17%, operating margins expanded 3 points to 23%, and free cash flow came in at 20% of revenue. 

HubSpot has also bought back nearly $1 billion of stock this year.

The most interesting part of the call was what customers are doing with AI. 

HubSpot is now the most popular CRM connector on both ChatGPT and Claude, and the usage isn't pulling customers away from the platform.

"They're not just pulling data out of HubSpot. They're doing more work in HubSpot," said Yamini Rangan (CEO, HubSpot). 

Customers who connect their AI tools to HubSpot retain better and upgrade paid seats at three times the rate of those who don't.

So the bear case says ChatGPT replaces your CRM. The data says ChatGPT users are coming in through the back door and doing more work in HubSpot.

AI is becoming a new revenue line too.

"Monthly credit consumption is up nearly 2.5x since the beginning of the year," Rangan said, as customers put HubSpot's agents to work on prospecting, support tickets and data enrichment.

But, the business faces some headwinds. 

Fourth-quarter growth is guided down to 12%. AI is changing how buyers discover software, which is hurting HubSpot's inbound demand. 

Big customers are taking longer to decide, and small ones are cutting costs. 

HubSpot is responding by moving to a model of seats plus usage credits, and it's willing to take a near-term hit to get there. 

We've seen this playbook before. When HubSpot cut seat prices in 2024, average selling prices fell at first, then the customer base grew and retention improved.

The valuation spread is the opportunity.

Look at the chart above. 

Forward earnings have climbed steadily for four years. The stock has gone the other way. 

Eventually the two lines have to converge.

HubSpot trades at 12x next year's earnings, the lowest in its history as a public company.

The free cash flow yield is 8.9%, an all-time high. That's a profitable, cash-generating software company with 311,000 customers, priced like a business in decline.

The bear case could still play out. 

Growth may slow further before the new pricing model kicks in, and the frontier labs are building their own agents. 

But at 12x earnings and a 9% free cash flow yield, the market is already pricing in the worst.

We bought HubSpot on Friday.

We'd rather own the platform that AI runs through than bet against it at these prices.

Lumida Curations

You can now view Lumida Curations in the Lumida Invest app.

Curated insights delivered on time. Download the app here.

Bridgewater's Research Machine

Bridgewater CEO Nir Bar Dea explains how the firm puts 1,200 people to work on questions the textbooks don't answer, and why every conclusion has to rest on real understanding rather than a good guess.

Did Netflix's New Fincher Film Use AI?

Jason argues that audiences care more about the story than how a film was made, while Lon says generative AI is just the next step from the algorithms animators and artists already rely on.

Meme

Not Subscribed Yet? Don’t miss out on future insights—subscribe to the newsletter now!

For real-time updates, follow us on: 

As Featured In

Disclaimer: Lumida Wealth Management LLC (“Lumida”) is located in New York, NY, and is an SEC registered investment adviser. SEC registration does not constitute an endorsement of the firm by the Commission nor does it indicate that the adviser has attained a particular level of skill or ability. Lumida only transacts business in those states in which it is registered, or qualifies for an exemption or exclusion from registration requirements. Any direct communication by Lumida with a prospective client will be conducted by a representative that is either registered or qualifies for an exemption or exclusion from registration in the state where the prospective client resides.

The information in this material has been obtained from sources believed to be reliable. While all reasonable care has been taken to ensure that the facts stated in this material are accurate and that the forecasts, opinions and expectations contained herein are fair and reasonable, Lumida, Inc. and Lumida Wealth Management LLC (collectively Lumida) make no representations or warranties whatsoever the completeness or accuracy of the material provided, except with respect to any disclosures relative to Lumida. Accordingly, no reliance should be placed on the accuracy, fairness or completeness of the information contained in this material. Any data discrepancies in this material could be the result of different calculations and/or adjustments. Lumida accepts no liability whatsoever for any loss arising from any use of this material or its contents, and neither Lumida nor any of its respective directors, officers or employees, shall be in any way responsible for the contents hereof, apart from the liabilities and responsibilities that may be imposed on them by the relevant regulatory authority in the jurisdiction in question, or the regulatory regime thereunder. Opinions, forecasts or projections contained in this material represent Lumida’s current opinions or judgment as of the day of the material only and are therefore subject to change without notice. Periodic updates may be provided on companies/industries based on company-specific developments or announcements, market conditions or any other publicly available information. There can be no assurance that future results or events will be consistent with any such opinions, forecasts or projections, which represent only one possible outcome. Furthermore, such opinions, forecasts or projections are subject to certain risks, uncertainties and assumptions that have not been verified, and future actual results or events could differ materially. The value of, or income from, any investments referred to in this material may fluctuate and/or be affected by changes in exchange rates. All pricing is indicative as of the close of market for the securities discussed, unless otherwise stated. Past performance is not indicative of future results. Accordingly, investors may receive back less than originally invested. This material is not intended as an offer or solicitation for the purchase or sale of any financial instrument. The opinions and recommendations herein do not take into account individual client circumstances, objectives, or needs and are not intended as recommendations of particular securities, financial instruments or strategies to particular clients. The recipients of this material must make their own independent decisions regarding any securities or financial instruments mentioned herein and should seek advice from such independent financial, legal, tax or other adviser as they deem necessary. Lumida may trade as a principal on the basis of its views and research, and it may also engage in transactions for its own account or for its clients’ accounts in a manner inconsistent with the views taken in this material, and Lumida is under no obligation to ensure that such other communication is brought to the attention of any recipient of this material. Others within Lumida may take views that are inconsistent with those taken in this material. Employees of Lumida not involved in the preparation of this material may have investments in the financial instruments or securities (or derivatives of such financial instruments or securities) mentioned in this material and may trade them in ways different from those discussed in this material. This material is not an advertisement for or marketing of any issuer, its products or services, or its securities in any jurisdiction.