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

Macro: Why the Consumer Won't Break; Why Are Yields So High?

Markets: When's the Buying Opportunity?; AI Infrastructure Keeps Getting Bigger; The AI Power Trade on Sale; Meta's Muse Moment; Most Hated Mag-7

Lumida Curations: Satya Nadella on AI Safety; Mark Zuckerberg on AI and Metaverse; Dylan Patel on Rising GPU Costs

Last week, we brought together a small group of family offices, investors, founders, and operators in New York for a cocktail party.

I sat down with Michael Parekh (longtime technology investor, and ex-Goldman Sachs partner who lead on Internet).

We discussed the demand for memory, China, and DroneTech. Listen to the conversation here.

Thank you to everyone who joined. Michael, as always, was a remarkable guest.

We'll host more of these in the coming months. Stay tuned.

Steve Jobs was a genius - I often play his ‘Connecting the Dots’ speech for my kids. They are way too young to understand it. 

But, they will remember one day when I am not around that I used to play it for them and hopefully they will appreciate the lessons.

Steve Jobs said that after seeing the Graphical User Interface at Xerox PARC, he knew this would be the future of computing.

He says it took him 10 minutes to realize it.

Steve came from the world before color GUIs. 

Before the mouse was normal. 

By the time he died, computers had transformed how the world worked.

I am starting to wonder if AI agents could be as consequential as the GUI.

The GUI made computers usable by ordinary people.

You no longer had to learn commands. 

You pointed, clicked, dragged and dropped, and opened.

It turned the computer from a specialist machine into a universal tool.

In the new world, the Computers that Steve Jobs brought into the world start to recede into the background.

To paraphrase Doc Brown from Back to the Future, ‘We don’t need computers where we are going.’

The Agents come to the fore and they interact with you on the channel you prefer.

The personal assistants we see today from Muse, Grok Bot, and Instinct are only the beginning.

Personal drivers. (FSD already does this).

Personal doctors. (Telehealth getting closer.)

Personal concierges. (Instinct does this.)

Personal wealth managers. (Lumida is building this.)

Personal fitness coaches.

Personal tutors.

Personal CFOs.

The Agentic world is where an AI knows your goals, has access to the right tools, and does work on your behalf.

In 2023, Chat GPT had no awareness of real-time news. 

It had no search integration.

Three years later, we're seeing highly productive autonomous agents. 

And the agents are writing other agents and learning faster than most humans learn.

I don't see Super Intelligence on the near-term horizon... 

But, that's not the relevant question for whether we're about to see large-scale transformation.

Three years ago I was firmly in the 'We're in the Slow AI Take-off' camp. 

Now I believe we are entering a new period of acceleration driven by Agents.

The change is coming fast. 

The current tools are still abysmally slow (not enough compute).

Jobs saw the computer go from a hobbyist machine to the center of modern life over his lifetime.

What took two decades in the age of the GUI may take a few years in the age of AI.

There's a real chance that agents are a bigger deal than the modern computer.

BUT, We are Getting MidTwit AI not Super Intelligence

Now, I still believe we are in a MidTwit AGI world.

That means we have highly efficient, productive, competent AIs that can execute a broad range of tasks - and better than most humans.

I don’t see Super Intelligence yet - there’s no Einstein or Newton coming anytime soon.

That means we won’t solve Fusion power, Bioinformatics, or invent new Battery storage or fuel cell designs with a prompt anytime soon.

The obvious winner from all of this is memory and compute.

You never get there. You aren’t there yet.

If you think you are, you probably are not.

It’s asymptotic.

New capabilities and tools are coming online.

Those tools are better, faster, and smarter than the old tools.

Every workflow, every customer, every decision - it all needs an agent.

It also takes time to ramp and train agents.

Once you have that, you need the agents to interact with your team - and each other.

Agentic / Team orchestration is a whole different kind of build.

The main imperative is to become AI native — and challenge yourself and your team to re-imagine what is possible.

Agentic transformation starts with leading from the top.

It’s not something you can delegate.

How you run your staff meeting (for example) needs to change. Performance reviews. Recruiting. All of it.

Properly understood, AI should flip how you execute inside and out.

No one is ‘there’ yet - the surface area is wide if you have the correct scope which is everything - but some are running faster than others.

(I will share in next week’s newsletter roadmap that will help you get AI Native quickly.)

This week, Tekever, a Portuguese-British maker of AI-powered drones, raised $580 million at a $6.4 billion valuation.

The company says its drones have logged more than 50,000 operational flight hours in Ukraine since 2022. And, investors paid for the record.

We saw the same a week earlier, when Shield AI was reportedly raising at a valuation of at least $20 billion, roughly double its valuation at start of the year, and our entry price.

(Disclosure: Lumida is an investor in Shield AI).

A significant part of the valuation gain was driven by the success of its Hivemind technology used in Ukraine.

The pattern is clear: companies with real deployments are being repriced quickly.

Government budgets point the same way. The Pentagon's FY 2027 budget request proposes more than $74 billion for drones and counter-drone technology, triple the FY 2026 amount.

Funding for the Defense Autonomous Warfare Group is expected to rise from $225.9 million to as much as $54.6 billion. Even AI stocks would struggle to match that growth rate.

The conflicts in Ukraine and the Middle East have moved warfare away from expensive legacy hardware toward cheap, expendable autonomous systems.

When an inexpensive drone can threaten equipment worth millions, every military has to rethink what it buys.

We remain excited about defense tech.

In private markets, 2018 to 2021 was the crypto cycle, and 2023 to 2025 was the AI cycle.

The defense tech cycle is underway now.

If you'd like more details on accessing Defense Tech deals that have the potential to IPO, you can sign up here. (This isn't an offer to buy or sell securities.)

If you are an accredited investor or qualified purchaser, you can join our deals communication list through Lumida Deals. The windows on these private deals can be short, so joining beforehand matters.

(Disclosure: This communication is provided for informational purposes only and does not constitute an offer to sell or a solicitation of an offer to buy any security or interest in any fund or investment. Any such offer will be made only to eligible investors and only by means of definitive offering documents, which should be reviewed carefully in their entirety, and only in jurisdictions where permitted by law. Investments in private, early-stage companies are speculative and involve a high degree of risk, including the risk of illiquidity and the total loss of capital; such investments are suitable only for qualified investors who can bear these risks. Any statements regarding the company's business, strategy, or prospects are forward-looking, are based on information believed to be reliable but not independently verified, and are subject to change without notice. Past performance is not indicative of future results, and there is no guarantee that any investment objective will be achieved. Nothing herein should be construed as investment, legal, or tax advice; recipients should consult their own advisors before making any investment decision.)

If you expected higher rates and inflation to break US consumers, you may be another Michael Burry – “too early and still wrong”. 

Retail sales rose 8.2% YoY last week, well above the 5.3% average for 2025. It has grown at a higher rate than 2025’s average for 12 straight weeks now. 

The usual explanation is the labor market, and it's a fair one.

Layoffs remain low, and fewer people are collecting ongoing unemployment benefits. 

People, who have jobs, feel confident and can keep spending.

But there is a second driver of consumer demand that gets far less discussion. 

The Baby Boomer generation.

Boomers are the wealthiest generation in history, and their spending depends more on their balance sheets than on their paychecks.

Boomers hold $97.4 trillion, more than half, of the total household net worth of $185 trillion. Millennials, in contrast, hold $20.8 trillion. Dave Ramsey might accuse Avocado toast and daily Starbucks for the difference.

Boomers also own $35.2 trillion, or 55.0%, of household equities and mutual fund shares. 

They also own $20.6 trillion, or 41.0%, of household real estate, the largest share of any generation.

In other words, Boomers own the assets that have appreciated the most over the past decade, and that wealth supports their spending.

Higher rates also affect them differently than younger households.

Rising rates hurt borrowers, but most Boomers are savers. They hold $9.6 trillion in deposits and money market funds, the highest amongst all cohorts. 

At current short-term rates, that generates roughly $350 to $400 billion a year in interest income. 

For this group, each Fed hike is effectively a raise. Warsh may be the most generous man in America if you're over 65.

They have also paid down most of their debt.

Boomers now account for 21.0% of household liabilities, down from 58.0% in 1990. They hold only 20.0% of consumer credit and 18.0% of mortgages.

Many have paid off their homes, and many others locked in low mortgage rates during the pandemic.

Inflation works in their favor, too. When prices rise, home values and other real assets tend to rise with them. 

Inflation squeezes a young renter's paycheck, but it grows an older homeowner's equity. Same inflation, very different Thanksgiving dinner conversation.

And because many are retired, their spending is far less exposed to layoffs or a slowdown in hiring.

The result is that a large and growing share of US spending power is protected from the very forces that are supposed to weaken the consumer: higher rates, inflation, and the oil shock.

Much of that boomer spending goes to leisure and experiences.

Boomers have the time, the savings, and a bucket list, and it shows up in strong demand for cruises and premium travel. The buffet line on a cruise ship may be one of the most underrated economic indicators in America.

That is why we like names such as Travel + Leisure (TNL), Delta (DAL), and Norwegian Cruise Line (NCLH). Each sells experiences to the generation with the most money to spend.

If oil prices keep easing, as we expect, lower fuel costs would add another tailwind for airlines and cruise lines.

US10Y reached 5.17% this week - you will have to go back about 20 years to see the last time it reached these levels. 

Two things are pushing yields higher: oil and growth — and higher returns on capital and greater need for capital (see hyperscaler debt issuance).

Oil gets most of the headlines, but growth is doing most of the work.

The TIPS (Treasury Inflation-Protected Securities) market shows how much each of inflation and growth contributes to the yields.

The 10-year yield has two parts. One is the real yield, which is what TIPS pay. The other is compensation for expected inflation, measured by the gap between the regular 10-year yield and the TIPS yield.

The 10-year yield is up 94bps this year. The TIPS yield is up 84bps. That leaves only about 10bps from higher inflation expectations.

Higher oil prices have nudged inflation expectations up, but only slightly. The inflation gap sits at 2.34%, inside the 2.0% to 2.5% range it has held since 2022.

The rest of the move is real yields, and real yields rise when the economy is strong.

The September PMIs from S&P Global confirm that. The services PMI jumped to 58.7 from 56.5, its highest reading in nearly five years. The manufacturing PMI rose to 57.0, its highest in more than four years.

A large part of that strength comes from the AI buildout. Data centers need land, power, equipment, construction, and labor, and companies are spending heavily on all of them. 

When the economy grows this fast, bonds have to pay more to compete for capital.

However, we do expect US10Y to topping around here. That means financials as a sector are cheap, and other rate sensitive categories such as Utilities.

XLU has sold off as yields rose. Look at the two charts below. We'll cover our picks in the Markets section.

Markets had a good week, led by tech.

Small caps, however, lagged, with the Russell 2000 down 0.8%. Many of these names carry heavy debt loads and need to refinance, so they stay sensitive to higher rates.

The next big event on the calendar is the midterms, now about five weeks away.

Prediction markets have made up their minds.

On Polymarket, the odds of a Democratic sweep of both the House and the Senate have climbed to 62%. 

Add the 31% chance of a Republican Senate with a Democratic House, and the market puts the odds of Democrats taking the House at over 90%.

Markets are largely pricing this in. And with a Republican in the White House, even a Democratic sweep means divided government.

Markets have historically been fine with gridlock. Washington gets less done, and investors don't mind.

History also tells us how this part of the cycle usually plays out.

The chart below tracks market returns across the four-year presidential cycle going back to 1949. 

The second and third quarters of a midterm year are typically the weakest stretch of the cycle, with the S&P 500 down 2.5% on average and the Nasdaq down 6.6%.

Then the sweet spot arrives. 

From the fourth quarter of the midterm year through the second quarter of the pre-election year, the S&P 500 has gained 20.2% on average, and the Nasdaq 29.4%.

Notice the dotted lines for the current cycle. This cycle has run well ahead of the historical average, which is why the midterm-year weakness has felt milder than usual.

We expect choppiness to continue until mid-October. Once the election is out of the way, the historical pattern points to a strong run into next year.

Earnings season is another reason to be optimistic. Q3 reporting starts in mid-October with the big banks, and that should give markets something more useful to focus on than headlines and yields.

The setup looks familiar. In 2023, markets spent the summer worrying about rising yields, and the pressure finally broke in late October just as earnings season started.

This time, the earnings picture is even stronger. 

AI demand remains supply constrained, earnings growth last quarter was broad across the index, and the consumer is still spending.

We expect companies to deliver another solid quarter, with AI infrastructure again doing much of the heavy lifting.

So when is the buying opportunity? 

We believe markets should provide a good entry soon. 

We are bullish on semis as AI infrastructure demand continues to soar higher, and names within the sector have returned to reasonable valuations.

Notice how the P/E ratio for the index has collapsed while earnings have continued to move higher. P/E ratio tends to be mean-reverting, and this spread presents the opportunity. 

If you want to know whether AI demand is slowing, don't listen to what CEOs say. Look at what customers are paying for compute. 

Ornn's B200 Rental Index just hit an all-time high - this index tracks the hourly rental spot price of Nvidia’s Blackwell-based B200 AI chip.

Renting an Nvidia B200 chip now costs an average of $7.88 per GPU-hour, up 68% from a year ago. 

Rental prices for a chip usually fall as it ages and newer chips arrive. B200 prices are doing the opposite, which tells you demand for compute is still running ahead of supply.

That demand doesn't stop at the GPU. Every chip that gets deployed needs memory, storage, wafers, networking, power, and a data center to sit in. 

When compute is scarce, the entire flywheel stays busy.

If you think about it, this comes in the same month when Anthropic's CEO called for a slowdown, and the CEO of the second biggest AI company, and the world’s wealthiest man, agreed. 

Someone isn’t keeping their end of the deal. Or, is Zuckerberg going full throttle?

Even if Anthropic and OpenAI ease off, the rest of the market won't mind. 

One less bidder at the auction just means Google, Meta, China, and the neoclouds get more chips at a better price. Someone will always take Dario's seat at the table, and they'll probably thank him for it.

The biggest buyers are also spending more, not less.

The largest US hyperscalers are on track to spend about $806 billion in capex in 2026, and Goldman expects $1.1 trillion in 2027.

At the start of this year, analysts expected only $546 billion for 2026 and $630 billion for 2027. Estimates for 2027 have risen by roughly 75% in nine months.

(The screenshot comes from the feed on the Lumida Investment App - an endless feed of curated insights. That’s my daily dose of dopamine)

These hyperscalers (Google, Meta, Amazon) are the most cash-generative companies in the world, and their capex is what funds the flywheel. 

Their spending becomes revenue for chipmakers, who then buy more equipment and memory to keep up. 

As long as the hyperscalers keep spending, the cycle keeps turning.

AI investment in data centers, power, networking, and chips is projected to total $10.3 trillion from 2025 to 2032.

That works out to 3.63% of US GDP per year… and yet, the P/E for the S&P is not in bubble territory. Nvidia is cheaper than the S&P.

The AI buildout is more than three times the size of the dot-com fiber boom.

We continue to maintain a significant position in Nvidia and TSM, and have names like Micron and SK Hynix.

These companies sit at the core of the buildout, and are essentially the toll booth for all funds that flow in the flywheel. Yet, they trade at compelling valuations, with earnings growth surpassing 50%.

The buildout is still in its early innings, and the checks keep getting bigger.

Utilities have had a rough couple of months. The Utilities ETF (XLU) has fallen about 15% from its late-July peak.

The reason is higher rates. 

Utilities pay steady dividends, so they are often treated as bond proxies. When bond yields rise, utilities fall, because investors can earn a similar income from Treasuries with less risk.

That relationship works in both directions. As yields come down, the group that sold off because of higher rates should be among the first to recover. 

Valuations across the sector are now far more compelling than they were in the summer.

Within this group, our favorite is Vistra (VST).

Vistra is not a typical sleepy utility. It is one of the largest independent power producers in the US, with a fleet that spans nuclear, natural gas, and renewables. 

Its business is heavily exposed to Texas and the mid-Atlantic, two of the most active data center markets in the country.

That exposure matters because power is now the key constraint in the AI buildout. 

Jensen Huang has said that securing land, power, and shell for a data center can take two to three years. Chips can be manufactured faster than the grid can be expanded, and nobody has figured out how to run a data center on vibes.

Hyperscalers want power that is reliable, available around the clock, and ideally carbon-free. 

Vistra's nuclear fleet checks all three boxes, and its gas plants provide the flexible power the grid needs when demand spikes. As AI demand keeps rising, the companies that already own generation hold the scarce asset.

The market hasn't priced that in. Vistra's stock has fallen more than 35% from its all-time high. That has pushed the valuation down to 13.4x P/E NTM. 

Look at the chart below. The green line marks roughly 13.6x, the level where the stock has bottomed several times since 2024. Each time it reached this level, it rebounded.

The more important chart is earnings versus price.

The blue line is forward EPS, which now sits at $10.29, near its highs. The black line is the share price. Since the stock peaked last fall, earnings estimates have moved meaningfully higher, while the share price has fallen sharply.

Earnings are rising, and the price is falling. That gap is the opportunity, and it tends not to last.

The bear case? Power prices are becoming a political issue. As data centers push up electricity demand, households are seeing higher bills, and that is the kind of topic politicians love heading into a midterm. Regulatory pressure on power pricing is a real risk to watch.

Still, the demand for power isn't going away. We are getting a leading AI power producer at a valuation near the bottom of its recent range, with falling rates as a potential tailwind.

Mark paid over a billion dollars to hire Alexander Wang. He has proven the return on that investment many fold!

Meta has had a meteoric rise over the last month, gaining about 40% in the process and making a new all-time high.

What drove it? Muse. 

Muse can help manage your email, book travel, compare products, and make purchases with your approval. You give it a task, and it keeps working in the background after you close the app.

This won’t be the last personal assistant app. Wait for Google to launch Spark - and I’d expect the stock reacts favorably too.

Notable about Muse — the LLM powering it is so-so. But, that’s not the relevant test. The useful of the app matters more.

Anyone that’s hired a smart person that can’t execute knows what we are talking about :)

Meta has now gained about $500B in market cap during September, which, to put in perspective, is worth more than the market cap of Bank of America, Costco, and Netflix. 

Bet on Founders

But, if you think about it, Zuck really made the right move replacing Turing Award winner Yann Le Cun with Alexander Wang.

That was not an easy call on Zuck's part as it is a tacit admission of a wrong decision.

But, that's life. Zuck fixed it rather than manage perception - what most corporations would have done.

Wang's parents were immigrants who worked at Los Alamos.

He went to MIT and dropped out of college at 19.

He joined Y Combinator.

He became the youngest self-made billionaire at 21 with Scale AI.

He's under 30 and putting Amazon on notice:

"I'm very excited to announce today that Walmart, Best Buy, Gap, Sephora, Wayfair, and many more are integrating with Muse to enable new shopping experiences for their users."

What an incredible adventure.

Here’s an interesting Non-Consensus Strategy.

Sheel Mohnot shared a strategy this week, where he backtested buying the most hated Mag-7 stocks ranked by 3, 6, and 12-month returns, rebalanced monthly. 

The strategy outperformed equal-weighted mags. 

We wanted to test it ourselves on a longer horizon, starting from 2015. 

Here are the findings:

(1) It works on a longer time frame going back to 2015.

(2) Results from 2015 beat equal-weighted Mag 7.

(3) The effect is bigger than 'owning Mag 7'

We have added this strategy to the Lumida Invest app, and you can now follow it to keep track of its performance. By this year’s end, we will have AI agents, helping you deploy and manage your portfolio on 15+

Lumida exclusive strategies. Download the app today.  

Here are the charts comparing Sheel’s strategy to SPY and Equal weighted Mag-7.

What’s the drawback with this strategy?

The Mag 7 of tomorrow aren’t the Mag 7 today. Nvidia was not part of the Mag 7 10 years ago. Should Broadcom be a part of this set? How about Taiwan Semiconductor? Anthropic?

Still, these are useful tools to generate ideas.

We expect that in a few months you’ll be able to subscribe to tactics like this in the Lumida Invest app.

Note: Backtest excludes transaction costs.

You can now view Lumida Curations at the Lumida Invest App. Curated insights delivered on time. Download the app here. 

Satya Nadella argues that the industry should treat AI safety as an engineering problem it can solve now, pointing to the joint safety board Microsoft and OpenAI have quietly run for years as proof that the model already works.

View Curation

Mark Zuckerberg says Meta expected the metaverse to come before advanced AI, and explains why Reality Labs now builds around AI as the bigger and more immediate opportunity, much as Apple shelved the iPad when it realized the iPhone was the bigger prize.

View Curation

Dylan Patel argues that existing GPUs are becoming more valuable: token prices are falling more slowly than the number of tokens each chip can produce is rising, so every GPU generates more revenue as models get more efficient.

View Curation

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

For real-time updates, follow us on: 

X | Telegram | Youtube | TikTok | News | Ram’s X | Lumida Health | Lumida Tax | Lumida Memes

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 transact 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.