Pinned Tweet
SALE LIVE: 20% off every Vanquish plan.
Pass Rapid Express in a single trade, or take Rapid Elite for an end-of-day drawdown.
10K Rapid Express: $63.20 (was $79)
10K Rapid Elite: $79.20 (was $99)
Keep 100% of your profits on both.
Use code VANQ through Nov 22.
JENSEN WANTS HIS $NVDA EMPIRE BACK
Jensen Huang announced today that he is willing to serve as CEO of Nvidia again. His statement is interesting because he frames his return as a way to serve others who have been calling on him to come back. He acknowledges his personal failures and accepts responsibility for the mess he created for the board to deal with.
He says he will cooperate with the investigation into said mess. But then he immediately goes on to describe the support he has received from employees, from investors, from customers, from business partners — the people closest to the business of operating NVIDIA through its most consequential period in history. And Jensen believes that they think he is the best person to lead them. That is a campaign to return to the CEO chair as soon as the board deigns to ask him to.
The letter also can be interpreted as one written to the U.S. government given how Jensen articulates the pride he has in immigrating to the U.S. 53 years ago and becoming a citizen and how Nvidia could only have been built in the U.S.
He then goes on to list various interests of the Board and concludes with the interest of America and how that interest should be weighed by the Board in its consideration of whether or not to bring Jensen back as CEO. Jensen’s statements will no doubt be received with skepticism by the government given his prior arguments in support of continued access to the Chinese market and the serious allegations of suspected Chinese intelligence agent activity at the company as well as intense scrutiny of the company’s export activities. To my reading, Jensen is attempting to regain the political capital he has lost with these letters and statements.
This earnings release from the largest AI compute company in the world comes at an inopportune time for a leadership change. Rather than focusing on near-term headwinds to growth and competitor success in the next generation of compute, investors will be focused on whether a CFO can make strategic decisions that Jensen Huang has been making for decades.
Support for Jensen’s return among customers, suppliers, and others is likely to alleviate overhang for investors, but the fact that the board reportedly has no intention to bring Jensen back as CEO, and instead has established a dedicated national security and export control oversight committee for Nvidia, suggests that reversing course now would undermine current attempts to restore credibility with the regulatory body that determines whether the company can continue to sell its most advanced chips internationally.
The biggest takeaway here, however, is that Jensen, for better or worse, is doing what he needs to do in order to get back into the CEO chair at NVIDIA. That is, he’s being presented to the employees, customers, investors and business partners that he needs to get back.
And it’s going to be up to the board to decide whether or not they can restore Jensen’s judgement to the point where they can continue to navigate the delicate web of export control and national security regulations that NVIDIA already operates under. And this letter is the start of that campaign. I expect we will hear a lot more from Jensen over the coming months.
1
9
704
FUNDS SOLD AI POWER IN Q3 BUT KEPT BUYING $META AND $NVDA
Q3 13F filings are in (positions as of Sep 30). Across ~6,400 filers (Reuters), 212 funds fully exited power and utility stocks, more than double the new positions. Data centers saw 138 exits vs 119 new positions.
Power and data centers, top 135 managers:
$GEV: 9 sold vs 1 bought, net -$603M (6 exited fully)
$NBIS: 4 sold vs 4 bought, net -$2B
$CRWV: 6 bought vs 3 sold, net -$571M (the sellers sold more)
$VRT: 5 sold vs 1 bought, net -$369M
$IREN: 4 sold vs 1 bought, net -$386M
Most bought:
$META: 21 buys vs 18 sold, net +$4.7B
$SKHY: 8 buys vs 0 sold, net +$4.2B
$SPCX: 11 buys vs 8 sold, net +$3.2B
$NVDA: 21 buys vs 11 sold, net +$3.2B
$AMZN: 22 buys vs 18 sold, net +$2.0B
Most sold:
$MU: 10 sold vs 6 bought, net -$5.7B
$SNDK: 7 sold vs 3 bought, net -$5.4B
$GOOGL: 38 sold vs 7 bought, net -$3.9B (10 exited fully)
$BE: 5 sold vs 4 bought, net -$2.6B
$AAPL: 10 sold vs 1 bought, net -$2.2B
Biggest single bet: Thrive Capital took $SPCX to 92% of its book.
Are funds early on selling AI power, or late?
5
8
788
$HOOD IS NO LONGER THE COMPANY YOU THINK IT IS
To become interested in Robinhood in 2026 you needs to look past the view of Robinhood as simply a speculative trading platform. The company has evolved into so much more since the wild times of the meme stock era and now operates as a brokerage, in retirement, in crypto, with prediction markets, banking, credit cards and even wealth management for customers.
In addition to the individual financial products for trading, banking, etc. offered by Robinhood, the company is building out a financial relationship with its huge customer base of young investors who are just starting to amass wealth. Robinhood is becoming a much bigger company than a mere brokerage.
A second thought Robinhood is trying to become the first financial service for a generation of wealth-accumulators. They’re not just trying to compete with Fidelity on individual financial products; they’re trying to become a customer’s primary financial service. HUGE opportunity embedded in that approach.
Despite a precipitous decline in crypto revenue (historically a large source of transaction revenue for the company), the rest of the business performed wonderfully in terms of transaction revenue growth. Indeed, prediction markets have now become a larger source of revenue for the company than crypto, options and equities trading all continue to produce meaningful transaction revenues and high growth rates in equities is more than offset by growth in options.
Prediction markets have replaced crypto as the largest revenue generator for the company, followed by options and equities with large transaction revenues in both areas. The company continues to add subscription, banking and other financial services to new areas of revenue for the company.
This decreasing dependence on the highs and lows of specific asset classes, however, does not necessarily mean Robinhood is a defense stock. For the most part, the company is still depending on there to be trading activity and associated sentiment shifts.
The real opportunity for Robinhood is as these financial products mature to the point where it can collect more on customers’ money sitting in their accounts than they can collecting on a margin of a trade.
The powerful incorporation of behavioral psychology to the way in which Robinhood develops products to serve its customers, bringing in money for them to invest and disincentivizing their transfer to other financial services providers.
By introducing Gold credit card with rewarding status levels to its customer base Robinhood attracts new customers. Later, once customers are on the platform subscription models introduce customers to even more financial benefits. Incentivizing long-term deposits of customers in their retirement accounts completes the build-up of customer assets on Robinhood’s platform. And, very importantly, the company’s platform to invest in and manage money in is extremely simple to use.
Also important is how these products interact with each other and how they can progressively add more value to the financial relationship that Robinhood is building with its customers. A customer that joins Robinhood because of a credit card offer (Robinhood Gold) might eventually establish a pattern of recurring investments or even transfer retirement accounts to the platform. Of course, none of these customers necessarily intended to become a customer of a primary financial platform.
So the major advantage Robinhood has in terms of psychology is all of these products interact to form a relationship with the customer. It would take a lot to get a customer who has transferred all their money into a retirement account to leave Robinhood for good.
While I can appreciate how the products of Robinhood are developed to generate a larger revenue stream and increase customer activity, it remains an open question how much margin the company can squeeze from all of this trading activity.
The increase in trading volume is far outpacing revenue per trade, and this will be a trend that must be monitored. Robinhood is attracting more and more trading activity, but they are earning less and less per trade due to tighter spreads, the shift to a more product-centric business model, and the use of incentives to drive customer usage of particular products.
In other words, the quantity of transactions is rising much faster than the corresponding increase in revenue per transaction. And management is carefully controlling and managing expenses in order to maximize profitability on this growth.
This leaves Robinhood to invest in many areas of financial services including banking, credit cards, international expansion and many new financial products – all of which will take considerable time to generate adequate returns for investors.
A good way to look at this is that the bear case for Robinhood is that the customer relationships that will generate so much more value in the future for trading are not generating the sufficient profitability yet for what investors are currently paying for the stock.
In terms of valuation, customer assets are likely to be a better indicator of Robinhood’s long-term earnings power than the revenue generated by their trading activity on a quarterly basis. Importantly, the amount of wealth that customers have accumulated in their Robinhood accounts continues to grow, even as their trading activity fluctuates between periods of high and low volumes.
Net deposits continue to roll into Robinhood even as overall trading activity is down, suggesting increasingly that customers are using the brokerage to accumulate assets rather than to trade frequently for short-term gain.
Of particular interest to me is how Robinhood’s interest-rate driven income from customer cash, Robinhood Margin, and other interest bearing assets on customers’ money will mitigate many of the other growth companies’ declining revenues in a higher interest-rate environment.
It is worth noting though that the growth in platform assets can stem from market appreciation in existing stocks held on the platform, as opposed to a rise in new customer deposits to the platform. Conversely, the assets under management on the platform could decline significantly in the face of market declines even as customers continue to deposit funds.
How Robinhood sustains asset growth will be critical to supporting its current valuation. Growth in assets under the platform and improving how much the company makes from that growth will be key to supporting current valuation multiples.
One thing that I find interesting about Robinhood is the way they are combining technology-enabled products with financial relationships that last for decades. However, as with most technologies-enabled products, there are risks to their business model, as I described in my list of potential issues above.
Thus, although customer growth translates into earnings growth for many tech companies, such is not necessarily the case here given, inter alia, decreasing transaction monetization, intense competition and, not least, growing uncertainty with respect to the extent to which certain of Robinhood’s very fast-growing products will be subject to significant and negative regulation.
As is usually the case with high growth stocks and an extremely high multiple, there are many risks here, and Robinhood’s payment for order flow, its emerging prediction markets, margin lending to customers and madcap expansion all will have the potential to kill the stock (at an extremely high multiple).
I’ll be focused on whether the company is able to keep adding assets under management, bring more customers to its wide array of financial products, and keep expenses down while only sacrificing marginally in terms of how much it is able to monetize each customer.
Longer-term, I am far more interested in whether they can get customers who open accounts to fund those accounts over time, start to use more of their financial products over time, and then eventually use the platform for retirement savings over time. If they can build that sort of relationship with customers, the future earning power of the company will look far different from the current perception of Robinhood as simply another online brokerage.
WHAT $NVDA NEEDS TO PROVE TO MARKET TOMORROW
I think the Nvidia earnings tomorrow will not be judged solely on the numbers, given the unusual circumstances of Jensen Huang’s resignation, Colette Kress acting as both interim CEO and CFO, and the independent board investigation into possible irregularities.
Given the very low valuation of 12x forward earnings ( lowest since 2011), I believe that a clear strategy for the company, a reassurance that the product roadmap is intact and some guidance on the leadership transition would be sufficient to instill confidence in the execution of the company’s strategy. The current low valuation does not provide a floor, given the unusual governance uncertainty.
What I’m looking for on the financial front is strong Q3 numbers of course, but also solid Q4 guidance and most importantly, visibility on the Vera Rubin mass production ramp for fiscal 2028. Hyperscaler CapEx has been revised up dramatically for the next couple of years with consensus estimating 100% spending growth in 2026 followed by 56% growth in 2027.
I don’t think anyone is doubting AI demand at present but rather how long such extraordinary growth can persist. A strong earnings report with solid Q4 guidance and associated margins (near 74% GMS) will certainly put a lot of pressure on the AI infrastructure spending at peak fundamental narrative.
My main questions here would be (i) are there still supply constraints limiting Nvidia shipments, (ii) Rubin production capacity scaling, and (iii) how much of such this extraordinary revenue opportunity actually translates into fiscal 2028 numbers.
I also expect questions around the regulatory environment to be abundant as the Commerce Department has announced restrictions on the export of advanced AI chips via remote access.
All such chips will now require licenses to be exported and additional licensing requirements will apply to exportations to customers in Singapore, Malaysia, Thailand, Indonesia and Brazil.
Importantly, these restrictions on the export of chips are distinct from a decrease in demand for chips by customers around the world. Indeed, restrictions on the ability of the largest and most successful AI provider in the world to service customers in certain geographic regions could result in a decline in revenues from international customers.
Importantly Nvidia will need to provide color around the potential impact to revenue from all areas of the international business, as well as specifically around Megaspeed. Importantly, investors will be unable to reasonably estimate these unknowns and as a result will continue to deduct an unknown amount from the current stock price as investors perceive this to be a risk that is currently unknown but inevitably will be disclosed.
Last but not least, a insight into how Nvidia will continue to fund the AI ecosystem under the Kress regime would be welcome. This includes their take on so called “circular” financing of neoclouds and related infrastructure from emerging AI companies.
So far, Nvidia’s growth depends on spending by ‘smaller’ customers, yet enormous cash-generative hyperscalers are committing to step up spending dramatically in the next few years. A highly successful earnings will more than likely result in a relief rally but in order to sustain the price increase for a longer period of time, much of that relief will need to be driven by a regaining of confidence in the exuded ability of Nvidia to sustain extraordinary earnings for an extended period of time from its current leadership and in its business relations with Washington, in particular.
WHAT IF EVERYTHING THE MARKET BELIEVES ABOUT $MU IS WRONG?
I continue to hold a very favorable view of the stock here as it is getting cheaper as the market starts to believe that the AI CapEx cycle may not be as durable as most believe.
And while the memory supply situation remains to be very tight for the time being, memory as a component in AI Infrastructure is becoming more important as a function of overall Infrastructure cost as customers seek to obtain the best memory at the lowest cost as AI begins to become more mainstream.
Here, Micron just reported $54.2B in quarterly revenue for the latest reporting period a stunning increase of 379% over the same period of the prior year as gross margin exploded to an astonishing 87% as the company managed to grow Cloud Memory revenue by 83% to $17.1B as the company’s HBM product, in particular, continues to command higher average price per gigabyte than its DRAM counterparts.
So, here, not only is the large increase in Memory spending by customers expected to continue for the near term, but Memory has become so critical to the function of overall AI Infrastructure that customers are beginning to treat Memory procurement as equally strategic as that of the GPU and associated Power infrastructure needed to support same.
Management at Micron has been hard at work signing 26 strategic customer agreements which are expected to represent greater than 35% of revenue for the company through 2030. Importantly, these agreements will translate into greater visibility into future revenue and profitability for the company.
Note that these agreements will, by necessity, have limitations to both the upside and downside for Micron over time, but in general, these types of agreements will provide a floor to pricing for the company in negative scenarios for memory prices as well as limit the potential for huge increases in pricing for the company in very positive scenarios for memory prices.
However, given the forward looking nature of HBM at 2027 and above, the vast majority of production for that year will have already been contracted at higher average price than was the case for 2026 HBM.
Remember that 83% gross margin Cloud Memory business is growing rapidly and increasing in terms of both shipments as well as average price per unit and thus there will be significant opportunity for improved profitability for the company as higher average price contracted products flow through the company’s numbers even in the absence of additional increases in overall memory prices.
In terms of the valuation of Micron, given its current price of ~7x EV/EBITDA over 2026 earnings (real cash generation is already being displayed), selloff driven downside is being priced into the stock which could result in substantial upside in valuation terms. Indeed, applying a highly pessimistic 3.5x EV/EBITDA multiple to Micron’s projected future earnings could result in 27% to 66% upside in stock price (note that these scenarios are highly dependent on future profitability that is already extremely high).
Importantly, bearish models and the current price of the stock are assuming substantial decline in the company’s future earnings. Meanwhile, the company is securing longer-term customer contracts and prices for its most strategic memory products are increasing, which makes current valuation so interesting.
Note that I am not extrapolating the current, extreme margins of the memory industry in infinity, as estimates for the company’s future revenue start to decline in a few years and eventually even margins will moderate down to industry averages by 2029.
On the other hand, there are several concerns for MU’s stock price that the market is currently pricing in. Higher interest rates to finance AI infrastructure are a concern as well as increasing political resistance to build out additional data center space.
A recent labor dispute at one of Micron’s Taiwan facilities that is currently under construction for approximately $4.5B (with the existing Taiwan facility not being impacted) has also recently received press coverage and adds to near term risks to further increase supply of memory in the near term.
However, tighter data center regulations are likely to be a negative near term for AI infrastructure only for a short period of time as MU believes that such restrictions will simply cause customers to build additional data center space in alternative geographies over a longer period of time. Such a scenario would certainly negatively impact earnings in the near term and reveal the highly sensitive nature of the stock to declining expectations.
However, Micron is transforming into a strategic memory leader that is moving from a highly cyclic memory manufacturer to a business where a greater and greater percentage of future production will be committed to contract with higher average pricing over time.
While there will certainly be cycles of increase and decrease in memory pricing, I expect the valuation to shift in a favorable direction over time as the percentage of future contracted demand increases.
$109,780.30 in payouts in one month. 🔥
Turbo absolutely ran it up in October, pulling $64,780.50 from his Rapid Express account alone.
He’s taking 100% of his profits home.
When's your turn? Use code "VANQ" to get 20% off through Nov 22.
5
6
1,069
WHY I THINK $NVDA IS STILL ON ITS WAY TO $10 TRILLION
First and foremost, I believe Nvidia will become a $10T company. Now, as with anything of that nature, it’s quite easy to keep investor fear going of losing so much value over the past two weeks.
But, to separate fact from fiction here, Jensen Huang was the founder, architect of strategy, the culture at Nvidia and major customer relationships. It’s huge news but in no way should this impact near term earnings power of the company he’s built.
Big difference. Leader leaves but Founder built massive earnings power. In fact, stock has fallen some 30% and yet infrastructure demand isn’t decreasing in any meaningful way. CUDA, networking, Rubin (production ramping), relationships w/ nearly all major AI infrastructure customers in the world and all this remains. Colette Kress (CFO) has been part of senior leadership since 2013 so massive uncertainty regarding who will permanently replace Jensen but in no way does that impact near term in my opinion.
To be clear, I’m not counting on a return to Nvidia dominance in AI compute. As I said, there’s increasingly serious competition from multiple companies including $AMD in its Helios designs, and from $GOOGL scaling up its own TPU compute silicon and building out its ecosystem of TPU clusters.
The growing AI compute industry will likely support the growth of multiple architectures, all capturing share of an exploding total opportunity. What I’m counting on is that the growing AI compute opportunity will be large enough that, despite increasing competition, Nvidia will keep growing.
This competition is very real, but it doesn’t mean the AI compute industry is stopping but that it’s becoming more normal and multidimensional. And in terms of valuation, it’s pretty clear that the recent Nvidia selloff created a purchasing opportunity at roughly 11x projected FY2028 earnings of $15.50 per share. Management has been guiding to roughly 70% revenue growth for that year, which would result in earnings of roughly 30% per year over the two years following, or 20x earnings in FY2029, or approximately $10T in market cap. That number (which again, isn’t a price target, and investors will undoubtedly assign a discount for uncertainty) is roughly where the fundamental growth in earnings of a leading AI compute company is tracking.
To be clear, we are evaluating a stock that has fallen nearly 30% in a relatively short period of time and is now trading at 11x projected FY2028 earnings of $15.50 per share for a company that management has outlined for investors to expect to grow greater than 70% in revenue for the year.
The forecasts for the company’s earnings growth have not deteriorated, they simply have not received the credit that they deserve given the significant change in leadership that is occurring. Therefore, if we assume that Nvidia’s earnings for FY2028 of $15.50 per share were to increase by 30% in the following year, or approximately $20.15 per share in earnings, then at 20x earnings the company would have a market capitalization of $10T.
Again, this is not to say that investors are suddenly going to price the stock at an extraordinary premium to where it currently trades. Rather, the earnings of the company will grow to a number where the market capitalization of the company will approach $10T in value as the scale of the infrastructure required to support AI computing grows exponentially.
I am extremely interested in how the company will execute on its earnings guidance for Q4 as well as its initial FY2028 growth projections for Rubin based off Jensen’s departure after 33 years.
Colette Kress, who has sat on the board of directors for 14 years and served as Senior Vice President of Sales, Service and Marketing for 14 years has apparently been earmarked by the board as Jensen’s permanent replacement for CEO.
While I recognize that Jensen built an enormous amount of value during his time as CEO of the company, it is possible that the company will maintain execution at the levels required to become a $10T company and indeed go on to become a $10T company provided that sufficient time is afforded to the new leadership to allow them to become familiar with all aspects of the business.
In the meantime, increased political resistance to construction of data centers, higher interest rates to service its increased debt and custom silicon designed to compete with the company’s GPUs all provide investors with excellent rationale for assigning a greater margin of safety to the stock than is currently embodied in the current stock price.
I continue to believe that the world is in the very, very early stages of building out the AI compute infrastructure required to support an immense number of end uses, and that accordingly, I believe that Nvidia remains one of the best positioned companies to profit from this amazing and enormous opportunity.
WHY I THINK 2027 COULD CHANGE EVERYTHING FOR $ONDS
I remain very bullish on the growth prospects of Ondas stock for 2027. The market views Ondas as having ‘purchased’ its growth via a series of large acquisitions – and thus largely ignores the R&D work being performed by Mgmt to build out core defense applications.
Here we have a company that in the latest Q delivered $167.5M in revenue and a pro forma organic growth rate of 64% and sees that rate increasing to 75%+ in Q4 with revenue in excess of $250M and a full year revenue number of at least $1.1B and greater than 40% organic growth. Importantly, all of this growth is expected to be realized without the need for Mgmt to complete any additional acquisitions and thus will all be generated by scaling out the already owned business units.
The economics of solutions to persistently monitor space for extended periods of time are getting very difficult to defend against ULTRA’s substantially lower cost structure compared to a whole range of traditional systems (not to be confused with snake systems like MQ-9).
When it comes to the intercept of the emerging threat of Shahed-class UAS, the cost of existing systems is hundreds of thousands of dollars per unit while Ondas’ IonStrike is designed to cost less than $50,000 per unit. So here, the potential to Ondas is not to win individual contracts here and there but to become a very large scaled supplier of low-cost, autonomous defense systems such as quantity, endurance, cost per mission. And with $835M of backlog currently on the books and another $214M of orders in Q3 alone, I expect that ULTRA and IonStrike programs will become very significant growth drivers as manufacturing capacity scales throughout 2027.
And when you put all these pieces together, the thought of what Ondas’s management is building is simply greater than the sum of the various defense companies they have bought up to acquire this revenue in the first place.
In terms of integration, GATE brings primary intercepted systems components, as well as precision munitions; Bron brings various critical components for GATE’s precursor systems and Stemme brings complete airframe production for ULTRA systems.
Ondas will now capture far greater margin on a much broader array of defense systems. With current adjusted gross margins of 34.8% (on the low end of the mix of precision strike and inefficient manufacturing), the various management teams have this thing punching 45% to 50% or so in adjusted gross margin over the next 12-24 months or so. With this kind of profit margin and projected revenue near or greater than $1.1B in 2027, the fully-earned earnings of Ondas’s amassed companies could be greatly different than what investors have been observing today.
I have a tremendous amount of respect for Ondas’s management team and the many decisions they have had to make in their effort to build out a comprehensive suite of autonomous defense products and services for military customers.
Execution risk remains substantial for many reasons including the fact that the Company continues to report (and expand) large losses of Adjusted EBITDA and that the Company’s cash has declined to $820 million.
However, Ondas has been becoming increasingly disciplined in its approach to considering M&A, and management now expects that the Company will become cash flow positive and will be reporting Adjusted EBITDA profitability at the Company level by Q3 2027.
Manufacturing capacity to support this growing revenue base is also being added. Importantly, the next several quarters will focus on converting a very large backlog of orders into shipments of both existing products as well as contracts for ULTRA and IonStrike that convert to production contracts. As has not yet been the case, I expect that at higher volumes of production, the Company will finally be able to achieve operating leverage.
I remain very bullish on Ondas’s long term prospects and believe that the Company is becoming something far greater than the simple aggregation of a number of individual companies that it has been for far too long. Market misunderstandings with respect to Ondas’s acquisition strategy are likely to be revealed in 2027 as to why management chose to buy the many companies that it has.
WHY I THINK THE MARKET IS GETTING $NBIS WRONG
The market is getting Nebius very wrong following earnings as it views delayed brings online of capacity as a change in demand to the company’s infrastructure. Actually the revenue is growing like crazy at 571% to $980.8M in the quarter.
Adjusted EBITDA comes in at $465.8M and most importantly, Nebius reports its first GAAP operating profit of $50.4M for the quarter. So yes, management lowered the guidance for 2026 to $3B to $3.2B and even more astonishingly introduced 2027 guidance of $12B to $13B. However, the biggest portion of lowered guidance for 2026 comes from delays at 3rd party facilities (like New Jersey).
So there is no cancellation of contracts by customers in none of the regions and thus no decrease in demand for Nebius’s infrastructure. Instead, the capacity will be brought online earlier in 2027 than expected and the company expects to reach more than $7B in ARR by February 28th vs. end of Decembder 31st of current year, which is a huge disappointment with management’s execution in the quarter. And yet, I am even more of a believer in the company’s long-term growth opportunity as it is preparing to quadruple in revenue in the year.
To add a bit more color to why I’m bullish here, the economics to the contracts are actually trending upwards while the stock is getting decimated. For Q3, Nebius signed new contracts for greater than $25M of annual contract value per MW of installed capacity.
The GPU pricing for new agreements signed in the quarter increased greater than 20% from the prior quarter’s signed pricing. Importantly, Nebius’s customers are paying greater than 100% of the associated CapEx for new contracts signed in the quarter compared to 50% to 60% in Q2 for comparable agreements.
More than 90% of the contracts signed in Q3 were signed with prepayments, and the payback period for comparable agreements signed in Q3 is approaching one year and 7 months of time to recover the CapEx for comparable agreements (prior to the impact of any prepayment receivable). Note that these are for new contracts signed in Q3, not for contracts signed in prior quarters.
It’s also worth noting that in addition to being a provider of GPU capacity to customers, Nebius is also becoming a software company, with agreements with Databricks, Datadog, Snowflake and Palantir, in addition to producing its own software through its Token Factory inference platform, which is quickly approaching multihundred-million-dollars in revenue.
This is where the market is getting $NBIS wrong following earnings. The selloff in the stock is treating delays in bringing on-line capacity as if demand for Nebius’ infrastructure is falling precipitously. While management did reduce 2026 revenue guidance to $3B-$3.2B and introduced 2027 guidance of $12B-$13B (below street), much of this revision is due to 3rd party facility delays that are pushing out the start of contracted capacity delivery to early 2027.
In other words, customers have not canceled and demand for Nebius’ infrastructure is still growing aggressively. Moreover, as noted above, management is forecasting that more than $7B in ARR will be recognized by February of next year, compared with prior guidance of $4.6B-$5.1B for all of 2026. In short, long-term growth opportunity is in no way deteriorating here. Rather, we are witnessing execution challenges with third-party facilities and a deterioration in near-term operating income.
This brings me back to my original thesis that Nebius can be a future megacap. Yes, the current operating metrics for the company for 2027 do not appear to reflect the true earning power of the underlying infrastructure being built out today. The market cap for Nebius is approximately $31B and trading at 2.5x the midpoint of the company’s own guidance for revenue for 2027 of $3.05B (or roughly 4x growth in revenue for the year). As noted above, on a GAAP basis, the company’s near-term earnings will be significantly impacted by the depreciation and interest expense of all of this new capital. In addition, there is a very real possibility of significant dilution as the company will be raising additional equity to fund growth.
Thus, I would recognize that the current stock price on a near-term earnings basis is very unattractive. However, on a long-term basis, I believe that there is significant potential for the stock to appreciate as the company moves beyond the initial build out of its large amount of infrastructure and into a period of high margin earning from the deployed capacity. In the short-term, much of the incremental margin from additional deployed capacity will flow through to operating income without the corresponding increase in expense.
Management expects to add more than 1 GW of new capacity per year and expects to improve pricing on its contracts with customers. In addition, the company is expected to go from being largely a provider of GPU-based infrastructure to also being a large provider of software to its customers, including its Token Factory-based inference platform that is expected to approach hundreds of millions of dollars in revenue.
Thus, while I would like to see much better execution around the delivery of the company’s scheduled build out of capacity and would like to have more visibility into the company’s planned financing, I remain a strong believer in the company’s ability to become a future megacap.
WHY I THINK THE MARKET IS WRONG ABOUT $RKLB
My initial take through the Rocket Lab earnings call is that the market is mistakenly viewing the Neutron delay in isolation and missing the very significant progress that has already been achieved prior to the launch of Neutron. Revenue for the quarter more than doubled to a record $267.8M with Space Systems revenue of approximately $215M or 80% of total revenue for the quarter. GAAP gross margin of 33.6% exceeded the company’s guidance of 29% to 31% for the quarter and the company reaffirmed guidance for Q4 revenue of $270M to $290M.
The company is already manufacturing satellites, optical communications equipment, propulsion systems and other key spacecraft components for commercial and governmental customers. In short, Rocket Lab is participating in the growth of the space economy.
I’d also note that a lot of people are looking at the Neutron delay through the lens of how long it will take to turn the $3.57B backlog of 117 launches and over 55 spacecraft in development and production into actual dollars. As noted above, only about a third of that total is expected to turn into revenue over the course of the next 12 months.
So it is frustrating but what really sticks out to me is that the contracts to add 8 more launches to the Neutron manifest of 16 at greater than $515M per launch were signed at a higher price ($70M per launch versus $55M) – and that Rocket Lab could have signed on for even more launches at that higher price. This is a company with increasing demand and improving pricing – and it has a significant delay in the launch of its next major product. That is very different from a company whose customers are losing confidence.
The even greater opportunity that Rocket Lab is developing is to become a space-based manufacturing company and launch service provider that can build spacecraft, launch them, and then operate the associated space-based infrastructure and services to generate recurring revenue.
The pending acquisition of Iridium will add a large established satellite communications network with more than 2.5M subscribers to the Space Systems segment, creating a large and growing source of recurring revenue from the associated communications services in addition to the Space Systems manufacturing and launch services.
The Company’s existing relationships with Defense customers are also gaining traction, with customers seeking to implement their own space-based infrastructure and services, where they can have greater control over the associated systems and services.
Rocket Lab’s financing situation is also stronger than it was earlier this year, with the Company holding approximately $4.7B in cash and securities and the acquisition bridge loan having been retired. Importantly, however, the flexibility that has been created has come via a significant equity raise, with the Company’s share count increasing by approximately 23% over the past year, and Iridium has substantial existing debt that will remain on the Company’s balance sheet following the close of the acquisition.
Operating cash flow even turned positive in the quarter for $7.2M, although free cash flow remained decidedly negative. In summary, I don’t think that Rocket Lab’s earnings are suddenly going to become profitable, but rather the Company is in the process of building out a large and much more diversified business, with improving future economics as the Company’s existing production programs scale, and the Neutron launch vehicle enters into commercial service.
The disappointment with Neutron is obvious as the first stage composite tank will require a refurbishment which will now push the inaugural flight to 2027 with no fixed launch date for the time being.
While the management reiterated that increased spend will be required until Neutron is flying regularly, production is advancing towards the recurrent production phase, adjusted EBITDA is expected to turn positive within six months post first flight, while sustainable free cash flow will take longer as it will depend on the rate of Rocket Lab’s reusable launch vehicles expansion.
Moreover, Rocket Lab’s stock price trades at a significant premium to its conventional aerospace peers. While I don’t think for a second that RKLB is cheap as it is trading at a significant premium to its peers’ stock prices and while there is plenty of competition from SpaceX alone and very Integration Challenges with Iridium’s satellite comms business, the market is definitely fixingate on the wrong things here (i.e. the timeline for Neutron’s launch) and missing the wood for the trees as the current Space Systems business is expanding very rapidly and commands large commitments from customers already while launch prices are moving up!
My only conviction here is that Peter Beck will translate all of this into very successful launches, ever-improving margins and, importantly, sustainable cash generation without too much further dilution for new investors. I remain very bullish on long-term potential of this company to become one of the space economy’s defining companies and today’s earnings actually strengthened that long-term thesis of mine even further than the corresponding drop in the stock price would suggest.
WHY $CRWV EARNINGS ARE SO BULLISH FOR $IREN $CIFR AND $NBIS
CoreWeave just did more to save the entire neocloud trade than I even care to recall during this AI drawdown. The biggest question that is keeping $IREN, $CIFR and $NBIS investors up at night is whether or not the industry is building out too much capacity in a market where demand and pricing could eventually collapse. CoreWeave’s most recent earnings report is quite bullish for the group as the company reported revenue growth of 168% to $3.7B, EBITDA of $2.3B and a $132.5B revenue backlog. Moreover, the company’s active power has more than doubled this year to more than 1.7 GW of power and the management team believes that qualified demand is running at ~10x available capacity through 2027.
I found it also very relevant that CoreWeave pricing for rentals has gone up 25% in July and an additional 10% in the subsequent month. Importantly, it is reported that shorter contracts (than the typical year) even generate more than $40M per MW per year in ‘annualized revenue’. I consider this also highly relevant for the Neoclouds as it appears to me that currently the industry is constrained by the speed at which high quality infrastructure can be rolled out rather than by a lack of willing and paying customers.
The same pricing for IREN: In the report CoreWeave disclosed pricing for a 5-year contract for a ~$22M per MW of GPU capital expenditures that are prepaid by the customer by 90% and pricing for a shorter contract of $40M per MW of GPU capital expenditures. CoreWeave also has more than 5 GW of power under contract to bring additional MW of infrastructure online.
As for Cipher, their main opportunity is to control scarce power and develop facilities that AI companies need to run their software. With their Barber Lake agreements now representing more than $9B in contracted lease revenue over twenty years, Cipher can take advantage of all this demand for infrastructure to run AI software for customers without taking on all of the risks of the GPU itself.
As for Nebius, their main opportunity for growth is to integrate the very scarce infrastructure that CoreWeave is building with an AI cloud and software platform for production inference of AI workloads.
This was also fascinating in terms of how the GPU contracts expire for CoreWeave (infrastructure still owned by CoreWeave as customers were paying to repay infrastructure financing). Ampere and Hopper GPU pricing had gone up 30% to 50% seq for CoreWeave in contracts that were expiring in 2027 (over 130,000 GPUs in total) which would generate $4B in annualized revenue at current pricing and $10B and above when the remaining 170,000 or so GPUs in 2028 contracts were included.
I also find it is very telling that CoreWeave is increasingly looking to offer services and software above and beyond simple rentals of GPUs as Forge and managed inference (now book $250M+ in booked ARR) are now being adopted by over 70% of new cloud customers. This creates huge opportunity to improve the economics for customers of infrastructure providers such as IREN, CIFR and NBIS while not forcing them to finance all of the incremental infrastructure spend which is why I find this report very insightful for these stocks as it underpins the massive AI infrastructure opportunity and shows how CoreWeave is turning demand into real cash flow for shareholders without destroying them in the process.
WHY THIS JENSEN HUANG SCANDAL COULD CHANGE EVERYTHING FOR $NVDA
I think dismissing everything happening with Nvidia CEO Jensen Huang as another round of FUD misses something important because part of what makes Nvidia Nvidia is Jensen himself and when CEO has become arguably the most influential voice in the entire AI industry, questions surrounding his judgment and potential exposure to foreign intelligence become much more consequential than an ordinary corporate scandal.
This is someone who has spent years building relationships with governments, shaping AI policy and convincing Washington that American companies should remain competitive in China, which is why I think investors need to separate their confidence in Nvidia's technology from the new risks surrounding the person leading it.
Where I think this becomes particularly significant is China's access to advanced computing, because Huang has been one of the most vocal advocates for allowing Nvidia to continue selling chips into that market while the woman at the center of these allegations was an executive at Megaspeed a major Nvidia customer that U.S. authorities have been investigating over suspected attempts to circumvent export restrictions.
Intelligence officials reportedly believe she had connections to Chinese intelligence, although that remains an allegation rather than an established public finding, and there is no confirmed evidence that Huang knowingly compromised sensitive information but even if the investigation ultimately clears him I think his ability to make the case for expanding Nvidia's access to China has become substantially more complicated particularly when lawmakers already skeptical of semiconductor exports now have another reason to question the industry's security practices.
That matters because the potential consequence isn't simply reputational damage, but a more restrictive political environment that could permanently reduce Nvidia's addressable market regardless of how strong demand remains.
And I think the broader issue is that Huang's influence extends well beyond Nvidia's products because he has effectively become the public face of the AI infrastructure buildout at a time when data centers, energy consumption and technology regulation are becoming increasingly political.
His acknowledgment of the affair and apology to his family and colleagues make this a legitimate question of executive judgment, while the independent board investigation the reported FBI inquiry and a shareholder lawsuit introduce governance and legal uncertainty that investors cannot simply dismiss because they like the stock.
At the same time, Nvidia's board has explicitly stated that it currently has no reason to believe company secrets or national security were compromised and I think that distinction is extremely important because speculation about the woman's intentions or the extent of any intelligence operation should not be confused with established facts.
My concern is that even without evidence of compromised technology, the controversy could weaken Huang's credibility with policymakers and provide ammunition to critics who already want tighter restrictions on AI infrastructure and semiconductor exports.
So I remain constructive on Nvidia's long-term fundamentals but I don't think a compelling valuation automatically means investors should ignore a developing situation that could affect the company's political influence, regulatory exposure and leadership credibility.
Nvidia reports earnings on November 17, followed by a congressional briefing requested for November 18, and I think those events become particularly important because investors will be looking for clarity on the board's investigation, the company's security controls and whether Washington's position on semiconductor exports is changing.
If the investigation finds no security compromise and policymakers ultimately separate Huang's personal conduct from Nvidia's business I would expect some of this uncertainty to fade and the valuation discount to narrow, but if the findings reveal deeper problems or accelerate restrictions on selling advanced chips into China, the consequences could last considerably longer than this news cycle.
I still think Nvidia is one of the strongest businesses in the AI economy, but that doesn't mean every risk surrounding its leadership is irrelevant, and understanding that distinction is ultimately what matters when deciding whether this selloff represents an opportunity.
New Rapid Express record just dropped 👀
Mauricio M. requested a $32,181.90 payout yesterday morning and got paid 18 minutes later.
Want in? Code VANQ gets you 20% off any account.
WHY $ASTS COULD BECOME A GIANT OF THE SPACE ECONOMY
After going through AST earnings, I am more convinced that we are watching the creation of a new layer of global communications infrastructure rather than simply another satellite connectivity company, because AST is combining proprietary technology in orbit with the distribution, spectrum and billing relationships of established mobile carriers.
Its successful AT&T terrestrial to satellite handover matters because ordinary smartphones need connectivity that works seamlessly within existing networks, while relationships with more than 60 operators representing over 3B subscribers give AST access to a customer base that would be enormously expensive to build independently, although that reach should never be confused with paying subscribers or a competitive advantage SpaceX cannot challenge.
I also think the financial distinction matters, because Q3 revenue grew to $42.3M from $14.7M but still came from gateway equipment and government programs rather than recognized SpaceMobile service revenue, while reaching management's $150M to $200M annual target now requires approximately $61M to $111M in Q4 alone.
The more consequential opportunity is management's expectation that revenue could approach $1B during the twelve months following commercial activation, which is very different from formal guidance for calendar 2027 and ultimately depends on carriers converting technical capability into actual customer usage and recurring payments.
Where I think the investment thesis becomes particularly interesting is that AST is moving from proving its satellites work toward solving the much harder problem of deploying enough capacity to make the service commercially useful because public orbital records show approximately twelve BlueBirds in orbit and reaching management's target of 45 by mid 2027 would require roughly thirty three additional successful deployments.
With BlueBirds 14 through 16 still awaiting their SpaceX launch and further deployments planned with Blue Origin in early 2027, the difference between manufacturing six satellites a month and actually operating a network capable of continuous coverage remains enormous, especially when management estimates that approximately 45 to 60 satellites are needed for more persistent service across priority markets.
What makes the technology roadmap compelling is the decision to introduce the next generation custom ASIC beginning with BlueBird 34 rather than BlueBird 50, because increasing processing capacity could allow each satellite to generate substantially more revenue without requiring a proportional increase in spacecraft, although the roughly tenfold bandwidth improvement management envisions relative to Block 1 still depends on available spectrum, beam capacity, interference management and ground infrastructure. At approximately $21M to $23M in direct material and launch costs per satellite before accounting for ground systems, operations and eventual replacement, the real economic breakthrough will come when greater orbital capacity produces higher paying traffic and better returns on every dollar invested.
The reason I increasingly believe AST could become much more important to the broader space economy is that the infrastructure being developed for consumer connectivity may also serve defense and sovereign communications markets, creating several potential revenue streams from the same underlying engineering investments.
The approximately $100M Space Force award associated with space based sensing is important because it demonstrates government interest in the capabilities of AST's large satellite arrays and onboard processing, although the award involves subcontracted development work and should not yet be treated as recurring operational revenue.
Japan potentially takes that opportunity further, with an approximately $938M government grant supporting the Rakuten joint venture in which AST owns 49% but I think investors need to understand that this is funding for the venture rather than unrestricted cash flowing to AST, particularly when the company also expects to contribute approximately $1.4B through preferred share subscriptions over several years.
The separate $1.2B government financing pipeline remains uncommitted, while Blacklift represents another attempt to control a critical part of the infrastructure stack by eventually reducing reliance on competing launch providers. I understand the strategic logic but the roughly $1.3B Firefly arrangement, potential share dilution and additional development spending introduce substantial execution risk, particularly because a rocket targeted for late 2028 cannot solve the immediate launch constraints facing the commercial network.
So I remain extremely constructive on AST becoming one of the defining companies of the space economy, but I think the strongest version of that thesis requires acknowledging that its strategic ambitions are expanding faster than its demonstrated cash generation, particularly when approximately $3.2B in reported cash includes roughly $435M of restricted funds, operating cash consumption exceeded $314M during the first nine months and Q4 capital expenditures alone are expected to reach $400M to $500M.
With the equity already valued at roughly $14B on a fully exchanged share basis at the November 9 close, my bullish scenario depends on launches and commercial activation accelerating through 2027 so that growing network utilization can support meaningful recurring revenue and eventually positive free cash flow, while my base case assumes a slower rollout extending monetization into 2028 and my bearish case involves further deployment setbacks, delayed carrier payments and financing that dilutes existing shareholders before the network reaches scale.
What I want to see now is independent confirmation of satellites reaching orbit and entering service, carriers activating paying customers, government milestones converting into collected revenue and expanding network capacity translating into improving cash generation rather than simply higher spending. If AST can demonstrate those developments through 2027 and into 2029,
I believe the market will increasingly recognize that the company is building something considerably more valuable than satellite phone coverage, potentially establishing an essential layer of commercial and sovereign communications infrastructure that becomes more strategically important as the network scales.
WHY $OSS COULD BE THE MOST OVERLOOKED AI STOCK RIGHT NOW
OSS could become one of most underappreciated Mini Monsters of AI trade because while everyone is focused on companies building increasingly powerful AI models considerably less attention is being paid to what happens when those models need to operate outside a data center.
An autonomous mining truck, military aircraft or counter drone system cannot always send information to the cloud and wait for instructions, particularly when connectivity is unreliable or decisions need to happen almost instantaneously which means increasingly sophisticated computing infrastructure has to operate directly inside these machines which is where OSS becomes interesting because the company specializes in packaging enormous computing capability into compact systems that can withstand extreme temperatures, vibration, physical constraints and limited power availability, essentially bringing data center performance into environments where conventional servers cannot reliably operate.
Once OSS becomes embedded inside a military aircraft, autonomous vehicle or other mission critical platform, replacing that computing architecture can require substantial engineering work and additional qualification creating meaningful switching costs that could support recurring production revenue, although competition from larger hardware manufacturers and increasingly standardized computing architectures means that advantage still has to be earned rather than assumed.
What makes me considerably more constructive today is that we're beginning to see evidence of this strategy moving beyond engineering prototypes and into actual production programs which I believe is the most important development underneath the company's recent financial results.
Since Mike Knowles became CEO in 2023, OSS has expanded from supporting one meaningful multiyear program to 14 while average order sizes have nearly tripled, and following the divestiture of Bressner, the remaining business is increasingly concentrated on the higher value computing systems where management believes it has the strongest competitive positioning.
Revenue grew approximately 36% to $12.6M in Q3, bringing nine-month growth to 48%, while management raised its full-year outlook for the third time to approximately 40% growth on the back of $43.6M in year-to-date bookings and a 1.46 ratio of bookings to revenue.
What I find especially compelling is that this momentum is no longer dependent on a single customer opportunity, with the P-8A Poseidon program generating more than $65M in cumulative contracted revenue, a medical imaging customer moving from prototypes into production and an autonomous robotics customer placing an initial $2.2M production order after successfully validating OSS's technology.
The robotics relationship could eventually represent $10M to $15M over five years, while an $8.4M defense award could develop into approximately $44M over four years, although those larger figures remain management estimates rather than committed backlog.
Today's additional $2.75M Army development award and the qualification of OSS's Gen5 4UP-MAX system with a major enterprise computing provider create further opportunities, but neither should be confused with established production revenue.
To me, the most interesting opportunity is actually the expanding middle tier of multimillion-dollar programs because developing a portfolio of customers that repeatedly order specialized computing infrastructure could prove much more valuable than relying on one enormous contract that may or may not materialize.
The part of the story I think investors need to understand is that OSS has already demonstrated it can win technically demanding programs, but the next phase requires proving that those engineering wins can translate into consistently attractive financial returns.
Ultimately what makes me believe OSS has genuine Mini Monster potential is the possibility that we're watching a relatively small engineering company establish itself inside a growing number of physical AI and defense platforms before those programs reach their full production potential, creating a situation where a handful of successful customer ramps could substantially change the company's earnings power.
My central scenario would be for revenue to approach approximately $70M by 2028, which could produce modest operating profitability if gross margins remain in the low 40% range and expenses grow more slowly than revenue while a stronger scenario approaching $90M or more with improving product mix could support double-digit operating margins and a meaningfully different valuation.
The bearish scenario is that production awards take longer to materialize, revenue struggles to move much beyond $50M to $55M and continued inventory investment or operating losses require financing, potentially offsetting much of the upside for existing shareholders.
I give greatest weight to the middle scenario today, but I think the upside deserves serious consideration because the progression from qualification to production has already been demonstrated in programs like P-8A, medical imaging and robotics, and additional opportunities across autonomous energy infrastructure, enterprise AI acceleration and Army computing systems could expand the number of platforms generating recurring orders.
What I want to see throughout 2027 is continued conversion of existing bookings into shipments, additional production orders from customers beyond the largest two, stronger gross margins as manufacturing volumes increase, positive operating cash flow as inventory converts into sales and evidence that the company can fund its growth without excessive dilution.
If those milestones materialize, I think the market could eventually recognize that OSS is developing something considerably more valuable than a small rugged computer business because it would be establishing an increasingly embedded position inside the infrastructure that allows artificial intelligence to operate in the physical world, where relatively modest revenue opportunities by industry standards can become transformative for a company of this size.
$ZETA COULD BE SETTING UP FOR A MONSTER RALLY AFTER EARNINGS
Zeta could be approaching one of its most important earnings catalysts of the year, although what makes this setup interesting to me is not simply the possibility of another strong quarter, but whether investors are about to recognize that the underlying business is becoming fundamentally more valuable than a traditional marketing technology company.
Zeta has now delivered 20 consecutive quarters of beating expectations and raising guidance, with revenue growing 44% to $443M last quarter and exceeding management's own forecast by $23M, while its upcoming Q3 revenue guidance of approximately $470M is almost identical to Wall Street's consensus which is why I believe the real catalyst is not whether Zeta beats revenue estimates again, but whether management provides enough evidence of accelerating organic demand, expanding customer economics and improving profitability to convince institutions that its earnings power heading into 2027 deserves a meaningfully different valuation.
What gives me increasing confidence in that possibility is what is happening underneath Zeta's reported growth, because although management expects approximately 39% to 40% revenue growth in Q3, that figure falls to 23% to 24% after excluding acquisitions and political candidate revenue, making it important to separate genuine platform demand from the contributions of Marigold, LiveIntent and the election cycle.
Interestingly, management only budgeted approximately $7M of political candidate revenue in Q3 and $8M in Q4, which means the election itself is a relatively small contributor to overall revenue and the more consequential question is whether Zeta can sustain its underlying growth as acquired businesses become part of the comparable base.
The customer economics give me a reason to believe it can, particularly with super scaled customers growing 17% to 197 while average annual revenue per customer increased another 17% to approximately $1.8M, effectively expanding the revenue generated by this important customer category through both new customer additions and higher spending within existing relationships.
What I find especially compelling is that customers adopting multiple use cases increased 90% while cross selling and upselling deals grew 43%, suggesting Zeta is increasingly expanding from individual marketing applications into a broader operating platform embedded across customer acquisition, retention and engagement.
If that expansion continues, the economic relationship becomes increasingly attractive because each additional workflow makes the platform more valuable to the customer while potentially increasing switching costs and revenue retention without requiring Zeta to acquire another customer which is ultimately how a software business develops more durable growth than its headline revenue figures initially suggest.
I actually think Athena could become the most important part of this entire investment thesis because Zeta is potentially moving from providing software that marketers operate themselves toward an intelligence layer that can actively help enterprises understand customers, identify opportunities and execute campaigns through natural language.
Within approximately 130 days of Athena's general availability, more than 40% of super scaled customers were already monthly active users, while management disclosed that roughly 20% of customers comprehensively adopting its AI capabilities accounted for approximately 70% of revenue with the most advanced adopters demonstrating materially stronger retention and revenue growth than customers still early in their adoption.
I would be careful about automatically treating that relationship as proof that Athena causes higher spending, since Zeta's largest and most sophisticated customers are naturally more likely to adopt advanced AI tools, but the opportunity becomes much more significant if management can demonstrate that comparable customer groups increase spending after adopting Athena rather than merely showing that its biggest customers use AI.
The underlying logic is compelling because reducing complex audience analysis, campaign development and customer targeting into conversational workflows can dramatically increase how frequently enterprises use Zeta's existing data and infrastructure, potentially allowing revenue per customer to expand faster than the cost of serving them.
Its Palantir partnership adds another dimension by connecting Zeta's customer intelligence with Foundry's broader enterprise operating environment, although I want to see those initial joint sales agreements translate into meaningful deployments and recognized revenue before assigning them substantial financial value.
As commerce increasingly moves toward AI agents, the companies that understand customer identity, intent and purchasing behavior could become more important, but only if their platforms remain relevant as traditional marketing channels evolve, which is why I think Athena's ability to turn proprietary data into measurable commercial outcomes matters considerably more than the number of AI features Zeta introduces.
The financial implications are where I think this setup becomes especially interesting because management is guiding toward approximately $405M in adjusted EBITDA and $255M in free cash flow for 2026, suggesting the company is beginning to demonstrate meaningful operating leverage even as it continues investing in its AI capabilities.
My bullish scenario is that Athena adoption continues accelerating, underlying revenue growth moves sustainably above the current low to mid 20% range and incremental margins improve as customers generate more revenue from infrastructure Zeta has already built, creating an increasingly attractive financial profile that could support higher earnings estimates and a stronger valuation into 2027.
The more conservative scenario is that Zeta delivers another respectable quarter but organic growth remains relatively unchanged while the bearish interpretation would be that acquisitions are masking weaker underlying momentum and AI adoption is failing to produce incremental profits after accounting for customer acquisition costs, stock compensation and dilution.
That's why I'll be watching the growth in super scaled customers, spending per account, retention among comparable Athena adoption cohorts, adjusted EBITDA margins relative to the approximately 24.5% Q3 target, free cash flow conversion and especially management's commentary about growth beyond the election cycle.
I believe this quarter has the potential to become a meaningful turning point for ZETA because the market has already seen that management knows how to grow revenue, but what it may be about to discover is that AI is allowing Zeta to extract substantially more economic value from the customers, data and infrastructure it already controls, and if that starts showing up consistently in organic growth, margins and cash flow, I think investors could begin valuing the company very differently heading into 2027.


















