VanquishTrader

$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.

THE FUTURE IS ON SALE RIGHT NOW Drawdowns from their highs: • $OKLO −88% • $OPEN −87% • $TE −85% • $AAOI −78% • $ASTS −76% • $QBTS −76% • $CRWV −75% • $NVTS −73% • $ONDS −73% • $AVAV −73% • $WULF −73% • $IREN −72% • $RKLB −65% • $IONQ −60%

WHY $ASTS BUILDING ITS OWN ROCKET COULD BE A GENIUS MOVE I understand why investors are questioning AST following the Blacklift announcement because committing more than $1.3 billion to develop an entirely new launch capability while simultaneously issuing additional shares creates another enormous engineering and financial obligation, but I think the market may be overlooking the strategic significance of what AST is actually trying to accomplish. The company already has third party launch agreements covering more than 100 BlueBird satellites through 2028, meaning Blacklift is not necessary to establish its initial commercial network, but rather represents a longer term effort to eliminate one of the few critical dependencies AST still does not control. SpaceX has historically been an important launch provider while simultaneously building Starlink's competing cellular network, creating an unusual situation where AST must rely partly on a competitor for access to the infrastructure required to expand its own business. The economics become particularly interesting when you consider that AST needs approximately 45 to 60 satellites to provide continuous coverage across its most important markets, because delayed deployment is not simply a scheduling inconvenience when adding the next group of satellites could determine how quickly the network transitions from intermittent connectivity into a consistently available commercial service. To me, Blacklift is therefore less about becoming a rocket manufacturer and more about transforming access to orbit from something AST must negotiate into something it can eventually control. What makes engineering strategy particularly compelling is that AST is designing Blacklift specifically around the requirements of its BlueBird constellation with 20 tons of planned launch capability, capacity for as many as five satellites per mission and reusable lower stages intended to support repeated deployments. Most launch providers design rockets to serve different customers with different payload requirements, whereas AST can potentially optimize an entire vehicle around the dimensions, deployment mechanisms and orbital destinations of its own unusually large satellites, which could become increasingly valuable as the company moves beyond building its initial constellation toward replenishing aging satellites and continuously adding network capacity. I also think licensing Firefly's Miranda and Vira propulsion technology is a strategically intelligent decision because AST is acquiring years of accumulated engine development and testing experience without having to recreate that work from scratch, with Miranda already completing more than 175 ground tests totaling over 4,200 seconds. That does not mean Blacklift itself is flight proven, particularly when integrating multiple stages, developing reusable flight systems and establishing reliable launch operations remain enormous challenges, but it potentially removes one of the most difficult components of the development process. The more important economic insight is that a dedicated launch capability could be valuable even without dramatically undercutting third party launch prices, because the ability to deploy satellites precisely when additional network capacity is needed may ultimately generate more value than the savings on an individual launch. Of course, the financial hurdle is substantial, and I would not call this a brilliant investment simply because vertical integration sounds attractive. AST is committing approximately $800 million upfront through $400 million in cash and 10 million to 14 million shares, followed by another $250 million for engines and $250 million for engineering support, with additional development and launch infrastructure costs potentially extending beyond the Firefly agreement. To put that commitment into perspective, spreading the initial $1.3 billion across 30 hypothetical lifetime launches would represent more than $43 million per mission before accounting for rocket manufacturing, operations, refurbishment or insurance, which illustrates why the investment cannot be justified by simply assuming internal launches will be cheaper. AST reported approximately $3.7 billion in pro forma liquidity following its July financing, but its existing satellite deployment program is already consuming substantial capital, making the opportunity cost of another major development program impossible to ignore. Where I think the additional upside becomes interesting is national security, particularly because AST has already disclosed more than $125 million in government awards, while a company capable of providing both resilient satellite communications and dedicated launch capacity could eventually offer defense customers greater operational independence and responsiveness. That is a potentially important competitive advantage, although existing government communications contracts do not establish that Blacklift itself will generate defense launch revenue. Ultimately I believe Blacklift could prove to be one of the most strategically intelligent investments AST has ever made because the company is potentially spending capital today to gain control over an infrastructure constraint that could become increasingly valuable as its network scales over next decade. The most compelling bullish scenario is not simply that AST launches satellites more cheaply, but that it eventually controls the entire process from manufacturing a BlueBird to placing it into the precise orbit required to expand commercial capacity, while reducing exposure to competing launch providers and potentially strengthening its position with government customers seeking assured access to space. What I will be watching is whether AST successfully qualifies its propulsion systems, completes integrated vehicle testing, demonstrates orbital deployment and reusability, and eventually delivers multiple satellites per mission at a fully loadedd cost that justifies the investment, while its underlying cellular business continues converting network coverage into recurring commercial revenue. If those milestones materialize, I think investors could eventually look back at Blacklift as the moment AST stopped depending entirely on other companies to determine how quickly its network could grow, because the most valuable thing AST may be buying is not a rocket, but control over the timing of its own expansion.

$IREN JUST PROVED THE MARKET WRONG TODAY I think IREN's latest earnings exposed a fundamental disconnect between what investors are afraid of and what is actually happening underneath the AI infrastructure economy, because the stock fell roughly 17% ahead of the report amid concerns about financing, Microsoft deployment delays and the sustainability of AI spending, yet the underlying results showed little evidence that demand for computing capacity is deteriorating. AI Cloud revenue nearly tripled sequentially to $200 million, comprising $135 million in cloud services and $65 million in infrastructure lease revenue, while operating annualized revenue doubled to $1 billion and adjusted EBITDA margin expanded from 14% to 29%. More importantly, contracted revenue backlog reached $20.3 billion, with approximately two thirds supported by investment grade counterparties, including Microsoft and NVIDIA which matters because customer credit quality influences how cheaply those contracts can be financed. I recognize that approximately $6.3 billion represents remaining service obligations while another $14 billion relates to lease arrangements, much of which remains dependent on future deployments, but that is precisely why I think investors need to distinguish the risk of delivering contracted capacity from an actual collapse in demand for that capacity. The market is increasingly treating the two as interchangeable, even though they have very different implications for IREN's long term earning power. What makes the quarter particularly interesting to me is that IREN is demonstrating how constrained computing supply can improve the economics of the infrastructure already available, especially when customers become willing to finance a meaningful portion of the equipment themselves. Since quarter end, IREN signed a five year agreement generating approximately $22 million in annual revenue per megawatt with the customer prepaying 90% of associated GPU capital expenditure, alongside a shorter agreement priced around $40 million per megawatt, while older H100 capacity was contracted at twice its previous pricing. These agreements are not directly comparable because hardware generations, contract lengths and utilization assumptions differ, but collectively they suggest customers continue to place significant economic value on immediate access to reliable computing capacity. The more revealing detail is in management's own illustrative economics, which assume roughly $60 million in combined GPU and data center investment per megawatt against $20 million in annual revenue and $3 million in direct operating costs, producing an estimated cash payback of approximately 2.4 years with substantial customer prepayments versus 3.5 years without them. Those prepayments reduce the capital IREN must initially finance rather than creating additional profit, and the calculations still need to withstand corporate expenses, depreciation, financing costs and future GPU replacement, but they demonstrate why favorable customer financing terms can materially change the returns available from the same physical infrastructure. In an environment where capital availability is becoming increasingly restrictive, securing customers willing to fund their own computing requirements could become just as important as securing the computing contracts themselves. That brings me to what I believe is IREN's most underappreciated competitive advantage, which is its ownership of the land, power infrastructure and data centers supporting those contracts rather than a business model dependent on long term leases from third party developers. Management reported approximately $6 billion of unencumbered data center assets, including facilities under construction, and that ownership provides flexibility to refinance infrastructure separately, accommodate customer design changes and potentially reuse buildings across multiple generations of GPUs instead of allowing the underlying real estate economics to accrue primarily to a landlord. However, I would distinguish those long lived assets from IREN's broader power pipeline because approximately 100 megawatts of IT capacity is currently operating, while the company's much larger portfolio includes facilities that still require construction, energization or regulatory approvals, including Sweetwater's conditional participation in ERCOT's Batch Zero process. This distinction becomes especially important following the midterms because greater restrictions on new grid connections could increase the scarcity value of infrastructure that is already operational while simultaneously delaying IREN's own expansion. I also think management's decision to moderate its spending ambitions reflects an increasingly rational approach to capital allocation, although the financial constraints are real because approximately $15 billion of planned fiscal year spending remains, with $9 billion supported by cash and committed funding sources and another $6 billion dependent on additional financing. The more than $4 billion operational ARR target is funded but has shifted into January or potentially February following Microsoft's requested design changes and supplier delays, which reinforces why owning infrastructure creates valuable flexibility without eliminating execution risk. Ultimately, my conviction is that IREN's earnings may have marked an important turning point in how investors should value the business, although I am not going to confuse impressive commercial momentum with evidence that shareholders are already earning attractive returns. At an equity valuation of roughly $8 billion following the initial earnings reaction, the opportunity looks increasingly interesting relative to more than $20 billion in contracted future revenue, but those contracts still need to support approximately $10 billion of debt, substantial remaining capital expenditures and the possibility of further dilution. The financial statements also make clear why the market remains skeptical, with a $263 million quarterly net loss and $5.1 billion in investing cash outflows while the $529 million operating cash inflow included a $645 million increase in deferred revenue, meaning customer advances remain an important source of funding rather than proof of sustainable operating cash generation. My bullish scenario depends on the remaining Horizon facilities receiving customer acceptance, operational ARR converting into recognized revenue and IREN securing additional financing at returns that comfortably exceed its cost of capital, while my base case allows for slower deployments and more conservative expansion, and my bearish case involves falling GPU rental rates, prolonged construction delays or increasingly expensive funding that forces material dilution. What I will be watching most closely is the actual conversion of contracted capacity into collected revenue, the financing terms for the next expansion phase, progress through ERCOT's approval process and whether recurring cash generation improves after accounting for interest, depreciation and future equipment investment. I believe the market may be underestimating how valuable scarce, usable AI infrastructure is becoming, but the opportunity for shareholders only becomes exceptional if IREN can translate that scarcity into durable cash flows without surrendering too much of the economics to lenders or new investors.

WHY $QCOM COULD BE THE BIGGEST AI WINNER OF 2027 2027 could be year investors finally stop looking at Qualcomm as a smartphone semiconductor company and start recognizing that Qualcomm is becoming one of the most important suppliers of computing infrastructure for the AI economy particularly because its earnings revealed two very different businesses moving in opposite directions. While handset revenue declined 26% during the quarter, automotive revenue grew 89% and management reiterated expectations for non handset revenue to grow more than 60% in fiscal 2027 supported by a data center business expected to generate $5 billion in revenue compared with Qualcomm's $43.1 billion of total revenue in fiscal 2026. What makes that interesting is that management already expects more than $650 million of data center revenue in the December quarter and an annualized revenue run rate of approximately $7 billion exiting fiscal 2027 which means this transformation should begin appearing in actual financial results over the coming quarters rather than remaining another long term semiconductor opportunity. The important distinction is that most of the initial $5 billion is expected to come from custom AI accelerator silicon for two hyperscalers and associated connectivity products, while Qualcomm's proprietary accelerator and CPU platforms represent additional opportunities that management expects to become more meaningful in subsequent years. That sequencing is why I think the opportunity could be considerably more interesting than investors appreciate because Qualcomm is beginning to commercialize its AI infrastructure business before the entire product roadmap is available, potentially creating multiple overlapping growth drivers as the company moves beyond its historical dependence on smartphones. The deeper reason I am increasingly bullish on Qualcomm comes down to how I believe the economics of AI computing are changing as the industry moves from spending enormous amounts of money training models toward running those models continuously across millions of applications and increasingly sophisticated AI agents. Training tends to reward enormous computational throughput, but generating tokens during inference can become heavily constrained by how quickly a processor retrieves model weights and previously generated context from memory, particularly during the decoding stage where the cost of moving data can become just as important as the calculations themselves. That is precisely why I find Qualcomm's High Bandwidth Compute architecture so compelling because its approach places computation closer to memory, potentially reducing the amount of data that needs to move across increasingly expensive electrical connections while improving the number of useful tokens generated within a fixed power budget. Qualcomm is targeting approximately 133 TB/s of effective memory bandwidth per AI250 accelerator, representing an 18fold improvement over AI200, although that figure reflects the architectural benefits of near memory computation rather than directly comparable physical DRAM bandwidth which is why independent workload benchmarks will ultimately matter much more than headline specifications. The recent validation of HBC silicon is encouraging because Qualcomm has demonstrated working memory pathways and computation in actual hardware, but AI250 is still expected to begin sampling in 2027 and has not yet demonstrated production scale economics against NVIDIA or AMD. What makes the opportunity strategically important is that Qualcomm is increasingly assembling the pieces of a broader inference platform spanning accelerators, CPUs, advanced memory integration, high speed optical connectivity and software, potentially allowing it to capture more of the semiconductor content required to operate AI infrastructure while extending its existing expertise in power efficient computing from personal devices into the data center. Where I think the investment thesis becomes particularly compelling is that Qualcomm is already securing commercially meaningful relationships before the market has fully adjusted its expectations for what the company could become with Amazon's multigenerational custom silicon and optical connectivity collaboration representing an opportunity potentially worth up to $60 billion over ten years. That figure is a commercial ceiling rather than a guaranteed purchase commitment, and the agreement includes warrants allowing Amazon to acquire up to 25 million Qualcomm shares as purchasing milestones are achieved, which means the economics need to account for potential dilution alongside revenue growth. I also think it is important to recognize that custom silicon can produce substantial revenue without necessarily delivering the gross margins investors associate with NVIDIA, particularly when hyperscalers retain significant architectural control and negotiating leverage, although Qualcomm believes its custom programs can still improve operating profitability as production scales. This is ultimately why I believe Qualcomm has the potential to become one of defining AI winners of 2027 particularly because the market appears to be evaluating its near term smartphone and margin challenges more aggressively than the longer term implications of its expanding position across the AI computing ecosystem. With shares around $157 following the earnings reaction, Qualcomm trades at approximately 15 times its fiscal 2026 adjusted earnings, which I find interesting for a business preparing to generate billions of dollars in data center revenue while maintaining a profitable licensing operation that helps fund continued technology investment, although that valuation is only genuinely attractive if declining smartphone earnings do not overwhelm the contribution from newer businesses. My bullish scenario is not simply that Qualcomm reaches its $5 billion data center target, but that investors begin recognizing that this could be the foundation of a much larger computing platform as inference becomes increasingly important, with the potential for revenue growth and valuation expansion to reinforce one another before the full earnings opportunity becomes visible. What I want to see throughout 2027 is the initial $650 million quarterly data center revenue target converting into sustained sequential growth, the business approaching management's $7 billion annualized exit rate, customer programs expanding beyond the initial hyperscalers and independently measured evidence that HBC can deliver superior inference economics at commercially relevant scale. Equally important, I want to see gross margins stabilize, manufacturing execution remain on schedule and incremental AI revenue produce genuine operating leverage after accounting for research spending and customer incentives, because the opportunity would become substantially less attractive if Qualcomm delivered its revenue targets without improving shareholder earnings. My conviction comes from the belief that AI is evolving toward a world where computing efficiency, memory architecture and intelligent devices become increasingly important alongside raw processing power and Qualcomm is one of the few semiconductor companies with credible technology across all three, which creates the possibility that 2027 becomes the beginning of a much larger transformation rather than merely another year of AI enthusiasm.

Me managing my 0DTE positions into NFP, CPI, and earnings season.

THE BIGGEST WINNER OF THE MIDTERMS COULD BE $EOSE I think the most important thing investors are overlooking about Eos following the November 3 midterms is that the real opportunity has much less to do with Democrats winning control of Congress and far more to do with a fundamental change in who is expected to pay for the electricity infrastructure supporting the AI economy. Democratic control of Congress should reduce the risk of further legislative cuts to existing Section 45X manufacturing and Section 48E storage incentives, although those credits remain subject to eligibility requirements and President Trump still has veto authority, which is why I think the more meaningful development is happening at the state level. Texas had already begun restricting new data center interconnections and environmental permits under Greg Abbott, while Gina Hinojosa's victory introduces the possibility of a longer moratorium and greater local control over future development. What interests me is that both political parties have increasingly arrived at the same underlying conclusion, which is that communities should not have to absorb the electricity and infrastructure costs of supporting enormous AI campuses, effectively shifting more of the responsibility for securing reliable power onto the developers themselves. That changes the economics of data center construction because electricity is no longer simply something a developer purchases after building a facility, but increasingly something it must secure, finance and potentially generate before that facility becomes commercially viable, although a broader permitting moratorium could still delay projects regardless of how independently they are powered. This is where I think Eos could become considerably more important than the market currently appreciates because making data centers responsible for their own power creates demand for an entire electrical system rather than generation equipment alone. A gas turbine can provide continuous electricity and solar can reduce fuel consumption, but neither eliminates the need to manage sudden changes in AI workloads, periods of excess generation or interruptions that could leave extremely expensive computing infrastructure sitting idle, which is why storage becomes valuable when it allows developers to use generating assets more efficiently while maintaining greater control over power availability. The distinction is that batteries do not create electricity and cannot independently sustain a data center through an extended outage, so their economic value depends on how intelligently they are integrated with generation and whether the resulting savings in fuel, infrastructure costs and downtime exceed the cost of the storage system itself. That is why I find Eos's zinc technology particularly interesting for applications requiring eight or more hours of storage, especially given its nonflammable chemistry and domestic supply chain, although its roughly 78% demonstrated round trip efficiency means it must overcome an energy efficiency disadvantage relative to conventional lithium systems through better lifetime economics, safety or longer discharge duration. The planned Google and MN8 project illustrates that opportunity particularly well because it combines ten hours of Eos zinc storage with four hours of lithium storage and solar generation to support the PJM grid, demonstrating how these technologies can complement one another rather than compete for every application while the TURBINE-X agreement targeting integrated gas generation and up to 2 GWh of Eos storage brings that same architecture closer to the independent power systems AI developers may increasingly require, even though those targeted deployments are not yet binding production orders. What gives me conviction that this opportunity could become financially meaningful is that Eos is already seeing the shift appear in its commercial activity, with management indicating that approximately 32% of its $24.6 billion opportunity pipeline is connected to data centers and supporting infrastructure, representing roughly $7.9 billion in potential projects before considering how quickly the regulatory environment is changing. However I think it is critical to separate that pipeline from the approximately $765 million of preliminary September backlog because even contracted projects can experience financing and construction delays, while a large pipeline tells us relatively little about future profitability until customers actually commit capital and equipment is delivered. This is why Frontier Power USA could become such an important part of the business model, with approximately $263 million raised to support a platform targeting more than $1 billion of project investment capacity, because access to development financing can help convert qualified projects into real orders, although Eos holds a minority interest in the venture and its financing arrangements introduce concentration, dilution and related party risks that investors should not overlook. In fact, a single Cerberus financed project accounted for approximately 80% of Eos's second quarter revenue, which is a reminder that financing driven demand is not necessarily the same as diversified customer adoption. The manufacturing economics are equally important because preliminary third quarter revenue of $82 million to $83 million still came with GAAP gross margins around negative 54%, meaning Eos was spending approximately $154 on cost of revenue for every $100 of sales, so the planned expansion toward roughly 4 GWh of annual production capacity at Thorn Hill only becomes valuable if higher utilization, improved yields, lower material costs and manufacturing credits fundamentally improve those unit economics. Its Golden Dome defense award provides another potential source of credibility, but neither that award nor its approximately $344 million cash balance, which includes restricted cash and has been supported by customer deposits, tax credit collections and DOE loan advances, should be mistaken for evidence that the underlying business has become self sustaining. This is why I believe the midterms may have provided Eos with the catalyst its turnaround needed, although I view the election as something that could accelerate an existing structural opportunity rather than an event that suddenly makes the company profitable. The fundamental change I am watching is that as AI infrastructure becomes increasingly constrained by the availability of electricity, developers will have stronger incentives to invest in systems that provide dependable power on commercially acceptable timelines and I think Eos has a credible opportunity to become an important supplier within that emerging infrastructure architecture if its longer duration zinc technology can deliver an attractive total cost of ownership. What I need to see now is the commercial conversion of that opportunity through binding data center orders, financed projects progressing into construction, actual deployments under relationships such as TURBINE-X and MN8, and independent evidence that Eos can deliver the reliability and performance customers require at a competitive price. Equally important, I want to see Thorn Hill translate higher production volumes into positive GAAP gross margins and eventually recurring operating cash flow that does not depend on customer advances, government credit collections or additional borrowing, because a company can participate in an enormous market and still destroy shareholder value if it cannot manufacture its products profitably. My conviction ultimately comes from the belief that reliable electricity is becoming just as strategically important to the AI economy as the computing infrastructure consuming it, and if Eos can turn that changing relationship into repeat customer deployments, improving unit economics and sustainable cash generation, I think its long term opportunity could be significantly larger than what investors currently associate with a struggling battery manufacturer, but the real validation will come from the business becoming economically stronger rather than simply the market becoming more enthusiastic about its story.

WHY THE MIDTERMS COULD ACTUALLY BE BULLISH FOR $CRWV AND $IREN The market is making an important mistake with CoreWeave and IREN following midterms because a policy that slows how quickly new data centers can be built is fundamentally different from a deterioration in the demand for AI computing and those two outcomes can have very different implications for companies that already control valuable infrastructure. Gina Hinojosa has called for a moratorium until lawmakers establish stronger community protections, which I take seriously given the potential implications for projects awaiting approval but what investors seem to be overlooking is that Greg Abbott had already ordered ERCOT to audit pending grid connections and halted environmental permitting before the election while pushing developers to cover their own electricity and infrastructure costs. The political direction was already moving toward greater regulation regardless of who won, which means the real incremental risk is how long restrictions remain in place and what conditions developers must satisfy rather than the sudden disappearance of an otherwise unrestricted market. I would also be careful assuming every permitted facility is automatically protected because electrical interconnection, environmental approvals and construction authorizations are different things but I think the market is moving too quickly from legitimate concerns about development timelines toward an assumption that the economic opportunity itself has somehow deteriorated. What makes this particularly interesting to me is that electricity is increasingly becoming one of the most valuable inputs in the AI economy because hyperscalers can purchase additional GPUs much faster than developers can secure hundreds of megawatts of reliable power which means the value of infrastructure that can support computing today should rise as the time and cost required to bring competing capacity online increases. The distinction matters because an operating megawatt with available cooling, networking and computing equipment has an entirely different economic value from a megawatt that exists only in an announced development pipeline especially if customers are willing to pay more for immediately available capacity. IREN is particularly interesting in this environment because Childress is already operational and its first Horizon deployment has been accepted by Microsoft while Sweetwater 1 has an energized high voltage substation and an initial 300MW of gross capacity under construction creating an infrastructure advantage that becomes harder for competitors to replicate as permitting becomes more restrictive. However, I would not treat the entire 2GW Sweetwater development as secured operating capacity because its inclusion in ERCOT's Batch Zero process remains conditional, which is why I think the opportunity is really about the increasing value of IREN's existing infrastructure alongside the execution risk associated with developing its remaining capacity. CoreWeave presents a somewhat different opportunity because its advantage is less about owning the underlying electricity infrastructure and more about its ability to turn available computing capacity into contracted customer revenue which becomes increasingly important when new competitors face longer development timelines. Galaxy has already delivered 133MW of critical computing capacity at its Texas Helios campus to CoreWeave and has the next phase financed and under construction but the distinction is that Galaxy owns the physical infrastructure while CoreWeave accesses it through long-term leases, meaning some of the scarcity value naturally accrues to the property owner before CoreWeave captures anything through its cloud business. Thats where I think investors need to be more sophisticated because CoreWeave's roughly $104 billion reported revenue backlog provides meaningful visibility into customer demand but does not represent cash already earned while its 9.625% unsecured debt financing demonstrates how expensive development delays can become when significant capital has already been committed. IREN faces its own version of this challenge given the substantial spending required to deploy GPUs and its approximately $13.8 billion of disclosed contractual commitments at the end of June so I do not view either company as an automatic winner from tighter regulation but I do think IREN offers greater exposure to the appreciation of scarce physical infrastructure while CoreWeave offers more direct exposure to monetizing contracted AI computing demand. Ultimately, this is why I view the current selling as an opportunity to buy the fear because I believe the market is increasingly pricing these companies as though restrictions on future development automatically undermine the economics of infrastructure that is already operating, when a more constrained supply environment could actually strengthen the competitive position of companies capable of delivering computing capacity today. The financial question is whether the additional value created by scarce power and higher utilization can exceed the present value of delayed expansion, increased construction costs and financing obligations, particularly since fixed customer contracts may prevent existing operators from capturing higher market prices immediately. For IREN, I want to see continued customer acceptance of its Horizon deployments, progress at Sweetwater, improving cash generation and evidence that its development commitments can be funded without excessive dilution, while for CoreWeave I will be watching whether its substantial backlog continues converting into recognized revenue and whether new capacity generates enough cash to comfortably support lease payments, interest expense and ongoing GPU investment. I would reconsider my conviction if permitting restrictions begin materially delaying existing projects, financing costs overwhelm the returns on new deployments or declining GPU rental prices prevent either company from benefiting economically from tighter infrastructure supply but my broader view remains that as electricity becomes an increasingly scarce resource in the AI economy, the companies with access to usable power and the ability to convert that access into profitable computing capacity could become considerably more valuable than investors currently appreciate.

Your best day shouldn’t stop you from passing an evaluation. Rapid Express has no consistency rule. Hit your target in one session or spread it across several, your call.

THE MARKET IS COMPLETELY MISSING WHAT $NVTS IS BECOMING Investors are underestimating the significance of what Navitas just demonstrated in its Q3 earnings because the company is no longer simply talking about becoming an AI infrastructure supplier but is beginning to show that its entire business is transitioning toward technologies required to power the next generation of computing. Revenue increased 38% year over year to $14 million, high-power revenue grew more than 50% sequentially and management reported a book-to-bill ratio of approximately 2.0 with backlog extending into 2027 which becomes particularly interesting when you consider that Navitas is achieving this growth while deliberately eliminating the mobile business that historically represented most of its revenue but I think the much bigger opportunity comes from a fundamental change in how data centers value power efficiency because as AI infrastructure becomes increasingly constrained by electricity available at individual facilities reducing power conversion losses can effectively increase the amount of computing capacity those facilities can support. Even a two-percentage-point improvement in overall conversion efficiency represents approximately 20 kilowatts of additional electrical headroom for every megawatt supplied to the conversion chain before accounting for cooling and other constraints which means more efficient power delivery can create economic value beyond lower electricity bills and that is ultimately why I believe Navitas could become much more important to AI infrastructure than its current revenue would suggest because the company is positioning itself around an increasingly valuable relationship between electrical efficiency and computing capacity. What makes that opportunity particularly compelling is that Navitas is approaching the problem across multiple stages of power delivery, which I think creates the possibility of several overlapping growth opportunities rather than depending on one technological transition. The industry's movement toward 800V architectures is partly about reducing the current required to distribute increasingly large amounts of electricity, since transmitting the same power at a higher voltage reduces current and the resulting resistive losses, creating opportunities for Navitas's SiC and GaN technologies throughout the data center but the engineering problem becomes almost the opposite as electricity approaches the processor, because delivering several kilowatts at core voltages below one volt requires thousands of amps to travel extremely short distances without generating excessive heat or compromising voltage stability. This is why I think the Claros acquisition could become strategically important, as its integrated voltage regulation technology allows power conversion to move closer to the processor through vertical and embedded architectures, potentially reducing losses while improving power density and the ability to respond to rapidly changing processor demand. Navitas now has technologies addressing everything from grid-level conversion through higher-voltage distribution and ultimately power delivery directly beside the XPU while its collaboration with NVIDIA on 800V conversion demonstrates how these capabilities could fit into emerging AI infrastructure designs. The distinction I find particularly interesting is that conventional silicon replacement is already contributing revenue, more advanced conversion architectures are entering the backlog, and integrated regulation creates another opportunity as future processors become increasingly power-intensive, potentially allowing several stages of the industry's evolution to contribute to growth simultaneously. And I think this is where the commercial evidence becomes just as important as the technology, because Navitas is beginning to demonstrate meaningful customer demand even though the transition toward high-volume AI infrastructure production is still relatively early. Management's disclosure of an approximately one-million-unit GaN order for a high-density 800V application is particularly encouraging because it suggests that at least some of these emerging architectures are progressing beyond demonstrations, while the expectation that AI infrastructure will represent more than 40% of revenue exiting 2026 and could approximately triple during 2027 provides a potential path toward sustained growth. However, I would distinguish that progress from guaranteed production revenue, particularly because a reference design developed with NVIDIA does not automatically translate into a large commercial contract, and the reported AI infrastructure category includes energy and grid applications alongside data centers. Competition is also becoming considerably more serious, with onsemi's proposed acquisition of Ferric demonstrating how established semiconductor companies are investing heavily in the same integrated power delivery opportunity that Navitas is pursuing through Claros and while the strategic positioning is becoming more compelling, the financial challenge remains substantial because Q3's 40% non-GAAP gross margin was accompanied by a GAAP gross margin of 21.5%, while fourth-quarter guidance calls for approximately $16 million in revenue against $25 million in non-GAAP operating expenses following Claros. That means Navitas still needs several times its current quarterly revenue to reach sustainable profitability at today's margin structure, making successful customer qualification, production scale and operating leverage essential to the investment thesis. Ultimately, the reason I remain so bullish on $NVTS is that the investment thesis I began developing when I first bought shares around $2 is increasingly being supported by the direction of the underlying industry, particularly as AI computing requires substantially more power, traditional electrical architectures become less efficient at higher densities, and semiconductor innovation becomes increasingly important to how much computing capacity can actually be deployed. What makes the opportunity different today is that Navitas has largely completed the strategic repositioning away from consumer electronics, assembled a much broader power technology portfolio and begun demonstrating accelerating commercial demand across several markets that could expand simultaneously over the next several years. I do not think the company has already established a durable competitive advantage across the entire power delivery chain, and I certainly would not assume that its expanding addressable market automatically translates into proportional revenue growth but I believe the combination of its technological positioning, growing backlog and emerging customer commitments creates a credible opportunity to become a substantially larger business. What I will be watching most closely is whether those commitments convert into repeat production orders, whether Claros begins contributing meaningful revenue, whether gross margins improve as the business scales and whether management can turn accelerating growth into positive free cash flow without excessive dilution. Because the bigger picture is that the AI economy will not be defined exclusively by the companies producing the most powerful processors, but also by the companies making it economically and physically possible to operate those processors at scale and I believe Navitas has one of the more interesting opportunities to become an important part of that infrastructure.

WHY TODAY’S $AMD EARNINGS WILL PROVE THE MARKET HAS THIS ALL WRONG AMD is heading into earnings with what I believe is a much bigger opportunity than simply taking GPU market share from Nvidia because I think investors are underestimating how the economics of AI computing are fundamentally changing as the industry moves from building increasingly powerful models toward deploying intelligence across the real economy. For years $NVDA advantage extended well beyond having the fastest accelerators because its hardware, networking and software created an integrated computing environment that customers could deploy with predictable performance but as inference becomes a much larger component of AI infrastructure spending I think the competitive equation increasingly shifts toward how efficiently these systems can deliver useful computation at scale. Hyperscalers aren't looking for another company to replace Nvidia because they have an enormous economic incentive to develop multiple viable computing architectures reducing dependence on any single supplier while optimizing different workloads around performance, power consumption and total cost of ownership and this is where I think AMD becomes particularly interesting because it does not need to displace Nvidia to become extraordinarily valuable especially if overall demand for computing expands dramatically as AI agents begin executing increasingly complex tasks across millions of businesses. What AMD needs to establish is that it can become a credible, scalable and economically competitive alternative that customers are willing to deploy across multiple generations of infrastructure because that would position the company to capture a meaningful share of an expanding market rather than simply compete for a fixed percentage of todays GPU spending and what makes me confident in that possibility is that AMDs opportunity extends well beyond Instinct GPUs particularly when you consider how the relationship between CPUs, accelerators, networking and software changes as AI systems become more autonomous. A traditional model inference request may rely primarily on an accelerator to generate an answer but an autonomous agent increasingly needs to retrieve information, coordinate different applications, manage memory, execute tools and orchestrate multiple computational steps, creating additional demand for the general-purpose processing infrastructure surrounding those accelerators. Thats why I think EPYC is such an underappreciated component of AMD's long-term opportunity, because the same expansion in AI workloads that drives accelerator demand can also create substantial incremental demand for server CPUs across cloud and enterprise infrastructure by combining EPYC processors, Instinct accelerators, Pensando networking and ROCm into a common computing architecture, AMD is attempting to compete on the economics of an entire AI system rather than individual semiconductor specifications, which could make the company more strategically relevant to customers optimizing infrastructure at the rack level. The distinction that ultimately matters is whether this integration produces genuinely competitive costs per token and better customer economics, because selling more complete systems does not automatically create higher margins, particularly when Nvidia retains substantial software advantages and hyperscalers continue developing their own custom silicon and I think AMD's relationships with OpenAI, Meta and Anthropic are becoming increasingly important evidence that this transition is gaining traction, although the underlying financial arrangements deserve considerably more scrutiny than the headline deployment figures suggest. OpenAI and Meta have each outlined plans involving up to six gigawatts of AMD computing capacity, while Anthropic has committed to an initial Helios deployment framework that could represent more than $20 billion in expenditures if the contemplated capacity is fully procured which demonstrates that some of the world's most sophisticated AI companies are increasingly willing to build meaningful portions of their infrastructure around AMD but I would distinguish that strategic validation from shareholder economics because AMD issued OpenAI and Meta warrants covering up to 320 million shares combined representing approximately 19% of its current share base if fully exercised, while separately committing $1 billion to Meta's equity offering and up to $5 billion toward Anthropic. I think the more interesting possibility is that these initial agreements help establish AMD as a proven alternative, allowing subsequent infrastructure purchases to occur on increasingly attractive commercial terms without requiring comparable incentives. If that happens then the first generation of contracts could eventually prove far more valuable than their immediate profitability suggests, although that outcome depends on AMD establishing sufficient performance, software compatibility and execution reliability to become a preferred supplier on economic merit alone. Ultimately I believe today's earnings represent an important test of whether AMD is beginning to translate these strategic advantages into the kind of sustained financial performance necessary to become one of the world's most valuable companies, particularly as investors look beyond the approximately $13 billion in Q3 revenue and 56% adjusted gross margin that management guided for and begin evaluating the scale of the Helios ramp into 2027. What interests me most is whether EPYC can sustain exceptional growth while Instinct deployments accelerate, whether AMD can protect profitability as more revenue comes from complete computing systems and whether the substantial working capital required to support these deployments begins converting into increasingly meaningful free cash flow. The evidence I will be watching is whether customers continue expanding deployments without extraordinary financial incentives, whether independent workload testing confirms AMD's competitive economics and whether those advantages translate into sustained free cash flow growth per fully diluted share because ultimately, I do not think AMD needs to become the company producing the world's most powerful AI chips to become one of its most valuable businesses but it does need to become one of the companies that makes the enormous expansion in AI computing economically sustainable and I believe AMD has a credible opportunity to become exactly that.

$SPCX MAY HAVE JUST HANDED $ASTS ITS BIGGEST ADVANTAGE I think one of the biggest things investors are overlooking about the reported discussions between SpaceX and Boost Mobile is that SpaceX becoming a direct competitor to traditional wireless carriers could actually strengthen AST's strategic position because the most important question is no longer simply who can build the largest satellite constellation, but who the existing wireless industry ultimately wants to depend on. SpaceX has spent years positioning Starlink as a complementary service that allows mobile operators to extend connectivity beyond their terrestrial networks but the possibility of acquiring Boost or launching its own integrated wireless service fundamentally changes that relationship. Instead of simply providing satellite connectivity to carriers, SpaceX could eventually compete for the same subscribers, pricing power and customer relationships that AT&T, Verizon and T-Mobile have spent decades and hundreds of billions of dollars building and I think this is where AST becomes particularly interesting because its business model is built around the exact opposite proposition, allowing carriers to extend their existing networks into space while maintaining control over their customers. What investors may be underestimating is that strategic neutrality can become a competitive advantage in its own right when the alternative is helping finance a supplier that could eventually become your competitor. And this is where the Boost discussions become even more interesting because acquiring a wireless brand is not the same thing as acquiring the infrastructure necessary to operate a nationwide wireless network. EchoStar reported approximately 7.38 million wireless subscribers in the second quarter but Boost had already transitioned its customer traffic away from its own radio network and onto AT&T's infrastructure which means SpaceX could potentially acquire the customers without automatically securing unrestricted access to the network serving them. Whether the underlying agreements require AT&T or T-Mobile to approve a change in ownership remains dependent on contractual provisions that have not been publicly established but the broader negotiating dynamic is important because SpaceX cannot simply replicate the terrestrial coverage and distribution of largest carriers overnight. T-Mobile's CEO made that distinction particularly clear during its October earnings call when he indicated that the company saw no room for a SpaceX mobile virtual network arrangement even as its existing Starlink satellite partnership continues. I would not assume T-Mobile abandons Starlink or that every carrier automatically moves toward AST, especially given the technology-neutral satellite initiative established by all three major operators but I do think SpaceX's ambitions create a stronger incentive for carriers to maintain an independent alternative and the more credible AST becomes as that alternative, the more negotiating leverage those carriers potentially gain over SpaceX which could make AST strategically valuable even before satellite connectivity reaches mass adoption. What makes this more than a theoretical argument is that AST is already demonstrating how its architecture fits into the existing wireless ecosystem, particularly with the newly announced successful handoff between AT&T's terrestrial network and AST's satellites using Ericsson and Nokia infrastructure. I think the significance of that milestone is that AST is attempting to make satellite connectivity function as a natural extension of the carrier's existing network rather than requiring consumers to adopt an entirely separate wireless relationship which could reduce integration complexity while preserving the carrier's control over distribution, billing and customer experience but technological alignment only creates economic value if AST can deliver sufficient capacity at a competitive cost, and this is where I think investors need to be realistic about SpaceX's advantages. SpaceX disclosed approximately 650 dedicated mobile satellites earlier this year, compared with AST's 13 satellites in orbit as of August and its next-generation mobile constellation and spectrum position could eventually create a substantial capacity advantage. However, those next-generation SpaceX satellites also depend on successful Starship deployment, while AST still needs to move toward roughly 45 to 60 satellites for continuous coverage across selected markets, meaning both companies face meaningful execution requirements before their long-term capabilities are fully realized. The bigger financial distinction is that SpaceX's potential retail strategy would introduce customer acquisition costs, network access expenses and pricing competition that AST's wholesale model can largely avoid, although AST must still fund an enormously capital-intensive constellation and accept that its carrier partners will retain substantial bargaining power over how the resulting revenue is divided. Ultimately, I think the most bullish interpretation of these developments is that SpaceX may be making AST more strategically important to the wireless industry at precisely the moment satellite connectivity is becoming commercially viable, because the established carriers increasingly need a credible satellite alternative without necessarily wanting to create another nationwide competitor. That does not mean AST is guaranteed to capture the opportunity, and I would not confuse having powerful partners with generating attractive economics from those relationships, because AT&T and Verizon can still negotiate aggressively, use multiple satellite providers and capture much of the value themselves. What would strengthen my conviction is seeing AST translate its carrier relationships into expanding commercial commitments, consistently successful network handoffs, reliable satellite deployment, meaningful recurring connectivity revenue and ultimately evidence that its wholesale economics improve as utilization increases. Conversely if SpaceX secures favorable terrestrial access, scales its next-generation direct-to-cell network faster than AST or drives satellite connectivity pricing substantially lower before AST reaches commercial scale, the strategic alignment advantage could become far less valuable than investors expect but I think the broader opportunity is becoming increasingly clear because the satellite communications market may not ultimately reward whichever company has the most satellites but whichever companies can make those satellites economically indispensable to the existing wireless ecosystem and if AST can successfully execute on its constellation while remaining the independent infrastructure partner that carriers trust to extend their networks without competing for their customers I believe that positioning could become substantially more valuable than the market currently appreciates.

The AI supply chain reports this month. Chips: $AMD on Nov 3, $ARM on Nov 4, $NVDA on Nov 17 Networking: $ANET and $ALAB on Nov 3 Optics: $LITE on Nov 5 Servers: $SMCI on Nov 3, $DELL on Nov 24 Neoclouds: $IREN on Nov 5, $CRWV and $NBIS on Nov 10 Which one are you trading? Save the full calendar of key earnings and macro events to plan your next session.

MARKET IS MISSING THE BIGGEST PART OF $MRVL $115 BILLION DEAL The market is making a mistake with Marvell's $115 billion acquisition of $LITE by focusing almost entirely on the acquisition premium, shareholder dilution and the decision to move beyond its traditional fabless business model, while overlooking what Matt Murphy is actually trying to accomplish. This isnt simply Marvell buying another optical networking company to expand its product portfolio, because I think the bigger opportunity is that AI infrastructure is approaching a fundamental architectural transition where moving data efficiently becomes just as important as processing it. As bandwidth requirements increase, the practical reach of copper connectivity continues shrinking, while the power required to move information between increasingly powerful processors becomes a much larger constraint on system performance. That is what makes the transition toward optical connectivity so important, particularly as the industry moves beyond connecting data centers and racks toward bringing optics closer to the processors themselves and what makes the timing interesting is that Marvell had already raised its fiscal 2028 revenue outlook to $20 billion before announcing this transaction which tells me Murphy is making this acquisition from a position of accelerating demand rather than trying to compensate for weakness in the existing business. Where I think market may be underappreciating the strategic value is that Marvell is not simply acquiring optical intellectual property, but the specialized manufacturing capabilities required to turn that technology into commercially viable products at scale since Lumentum manufactures indium phosphide lasers that are becoming increasingly important for next-generation optical interconnects, and its new 240,000-square-foot North Carolina facility is designed to expand production of advanced lasers using six-inch wafers, with manufacturing expected to ramp in mid-2028. The important distinction is that developing an optical architecture and manufacturing the lasers required to support it are two very different challenges, because production yields, thermal performance, reliability and customer qualification determine how much of the theoretical manufacturing capacity can actually become usable supply. By combining Lumentum's laser manufacturing with Marvell's optical DSPs, silicon photonics, switching and custom silicon capabilities, Murphy is positioning the company to optimize more of the optical signal chain rather than relying on separately developed components but the real question is whether ownership can create advantages that commercial partnerships cannot particularly because the companies were already demonstrating interoperable optical systems before the acquisition and both Lumentum and Coherent have secured upstream indium phosphide substrate commitments. I think the opportunity is less about monopolizing laser supply and more about controlling the engineering process across components that will increasingly need to be designed together as optical connectivity moves deeper into AI infrastructure. That brings me to the financial concerns because I actually think the market has legitimate reasons to question the acquisition price, especially when Lumentum shareholders will receive 4.5 Marvell shares for every share they own and ultimately control approximately 33% of the combined company. The fixed exchange ratio also means the dollar value of the consideration changes with Marvell's stock price, so the original $1,132 implied offer is not a guaranteed cash payment. Management is projecting more than $1 billion in annual cost synergies and adjusted EPS accretion during the first year after closing, but neither figure tells us whether the transaction will ultimately generate enough incremental free cash flow to justify the equity being issued. There is also a meaningful change in the economics of the business because owning laser factories introduces additional capital expenditures, depreciation and manufacturing yield risks, although Lumentum's recent 47% GAAP gross margin demonstrates that specialized optical manufacturing can already generate attractive profitability under favorable conditions. Ultimately the reason I believe the market could be underestimating this transaction is that the next phase of AI infrastructure may create substantially more economic value for companies capable of delivering complete, optimized connectivity systems rather than individual components and Marvell is positioning itself to participate across a much larger portion of that opportunity but the acquisition only makes financial sense if combining these technologies produces measurable advantages in bandwidth, power efficiency, manufacturing yields and customer deployment timelines that would have been difficult to achieve independently. For me most important things to monitor will be whether Lumentum successfully ramps its new manufacturing capacity, whether Marvell secures additional optical design wins across scale-up and co-packaged optical architectures, whether customer relationships remain intact and whether the combined business ultimately delivers higher normalized free cash flow per share after accounting for dilution and manufacturing investment. If those outcomes materialize then I think investors could eventually look back on this transaction as another defining moment in Marvell's evolution, similar to how the strategic importance of Inphi became much clearer over time. Im not suggesting that paying $115 billion is automatically justified because optical connectivity is becoming increasingly important, but I do believe the market may be underestimating how valuable it could become to control both the semiconductor intelligence responsible for moving data and the specialized optical manufacturing required to move it at scale particularly in an AI economy where the ability to connect increasingly powerful compute systems efficiently could become one of the industry's most important competitive advantages.

WHY $PLTR COULD BECOME THE MOST VALUABLE COMPANY IN WORLD I think one of the biggest mistakes investors are making with Palantir is continuing to evaluate the company through the traditional enterprise software framework because what makes this opportunity so interesting to me is that AI could fundamentally change where value ultimately gets captured. For decades companies have purchased software to help employees perform their jobs more efficiently, but I think we are approaching a much bigger transition where software increasingly moves from assisting human decisions toward autonomously executing them across entire organizations which changes the opportunity from selling applications based on how many people use them toward capturing a portion of the productivity and economic output those applications generate, which is where I believe Palantir becomes particularly interesting. The biggest bottleneck in enterprise AI is no longer necessarily the intelligence of the underlying model, but its ability to understand how a business actually operates including the relationships between its assets, suppliers, employees, permissions and operational constraints and that institutional knowledge cannot simply be purchased through an API, which is why I think Ontology could become one of the most valuable layers of the AI economy by connecting intelligence to the actual operating environment of an organization and allowing decisions to translate into real actions. What makes this architecture particularly powerful is that every additional workflow integrated into Ontology potentially makes the surrounding system more valuable, because the relationships, permissions and operational decisions established through one application can become the foundation for automating additional processes across the organization. Over time that creates a compounding advantage where customers become increasingly dependent on the platform as more of their operations run through it, while the cost and complexity of replacing that infrastructure continue increasing where I think this becomes even more important as foundation models get cheaper and increasingly interchangeable, because lower intelligence costs should make thousands of additional workflows economical to automate, but every one of those agents still needs access to proprietary information, institutional context and a controlled way to execute decisions. Palantir doesn't need to develop smartest model in the world if it can become the operating environment underneath whichever models enterprises choose to deploy and the implications could extend well beyond traditional software budgets since a manufacturer increasing production without building another factory or an airline reducing costly maintenance downtime can generate enormous financial value, which creates an opportunity for Palantir to capture a meaningful portion of those improvements. The distinction is that creating value does not automatically mean capturing it, so the long-term opportunity ultimately depends on whether Palantir can prove that its platform delivers outcomes customers cannot easily replicate elsewhere and I think the financial results are already providing some of the strongest evidence that this model is working What makes that particularly interesting is how Palantir's business model could evolve as its bootcamp distribution strategy and AI forward deployed engineering capabilities reduce the time and specialized labor required to build additional applications, potentially allowing the company to combine the switching costs of deeply integrated enterprise infrastructure with the scalability of traditional software. The company really doesn't need to monopolize enterprise AI but it does need to demonstrate that its ability to connect fragmented systems and execute complex operational workflows remains sufficiently differentiated to support continued expansion and pricing power which is why I believe Palantir has a credible path toward becoming one of the world's most valuable companies although I think investors need to separate the magnitude of that opportunity from what the current valuation already assumes. Intelligence will become increasingly abundant so the scarce and valuable layer may be the infrastructure that allows thousands of AI agents to understand, coordinate and safely execute decisions across the real economy so if that transition plays out the way I expect then Palantir has the potential to become much more than another enterprise software company because it could become one of the foundational operating layers of the autonomous enterprise which is exactly why I believe it has a legitimate opportunity to become one of most valuable companies in the world.

Trick or treat? We're going with treat 🎃 We were going to end 50% off today, then remembered it's Halloween. Half-price evals are staying a little longer. Use code HALLOWEEN before Nov 2, 11:59 PM EST.

COULD $TSLA and $SPCX BECOME THE MOST VALUABLE COMPANY IN THE WORLD? I think the real question around a potential Tesla and SpaceX merger is whether common ownership improves AI infrastructure economics because their commercial ties are real but hardly transformative since Tesla booked $626M of Megapack revenue and $148M of gross profit from SpaceX through September while its $2B xAI investment became a SpaceX stake worth $2.87B representing less than 1% ownership and proposed AI software and robotics collaborations lack material disclosed revenue. A merger would eliminate those sales from consolidated revenue, while its real incremental benefit would be jointly financing chip capacity and preventing stranded infrastructure as demand shifts. If supply contracts can accomplish that, the merger premium should be minimal. That distinction matters with orbital compute, where SpaceX's reusable rockets and Starlink network offer genuine capabilities but commercial cost superiority remains unproven. Its prospectus envisions 100kW of compute power per metric ton, meaning $1,500/kg launch costs alone imply $15M per orbital MW before processors, spacecraft, solar arrays, radiators, radiation shielding, connectivity and replacement launches and independent modeling estimates orbital compute costs of $660M to $750M per MW over 20 years versus $230M to $300M terrestrially, since near continuous sunlight avoids grid expenses without eliminating the weight, heat and failure penalties of operating hardware in vacuum. Tesla's batteries can alleviate terrestrial power constraints, but they can already be sold to SpaceX, so what matters is whether integration lowers cost per useful token. Terafab offers the clearest potential for a structural advantage because combining logic, memory and advanced packaging could support cheaper, more secure silicon for Tesla's planned autonomous fleet and SpaceX's AI clusters. Yet the announced $16.8B first phase is an estimated investment, not proven wafer economics, and the prospectus warns Tesla and Intel need not remain involved while specific projects still require separate agreements. Before crediting a synergy, I want funded equipment orders, competitive yields and lower cost per usable chip than established foundries. The risk is especially asymmetric given SpaceX spent $40B on Super Intelligence capex in nine months, meaning a deal could simply move a capital intensive buildout onto Tesla's shareholder base. At October 30 prices, Tesla and SpaceX together were worth approximately $3.24T after eliminating Tesla's stake, still $1.73T below Nvidia. Even a generous standalone sum of the parts valuing Tesla autos, energy, unproven robotaxis and robotics at $400B, $200B, $300B and $200B, then SpaceX launch, Starlink and Super Intelligence at $200B, $800B and $1.1T, totals only $3.2T. At an aggressive 40 times sustainable free cash flow, $5T, $7T and $10T require $125B, $175B and $250B in annual FCF, implying $500B, $700B and $1T of revenue at 25% cash margins and roughly 35% operating margins, against $105B in combined revenue and negative $33B in FCF through September. At market parity Tesla holders would own roughly 44%, subject to an unknown exchange ratio and Musk's supervoting control, so I would only credit additional value after independently approved terms, competitive Terafab yields, orbital cost per token evidence and sustained cash flow after capex over the next 12 to 36 months.

Another record payout from Brian A. Last week, he cleared a Rapid eval and requested a $14.8K payout. Today, he did it again in just 22 hours. PAYOUT SPOTLIGHT - Trader: Brian A. - Plan: $50K Rapid Express - Cost: $125 (with discount) - Payout: $19,873.00 Want in? Code RAPID gets you 50% off.

WHY $INTC IS ON TRACK FOR ITS WORST TRADING DAY IN OVER 50 YEARS Intel is going through it today after Q4 revenue guidance of $15B to $16B missed the $17.1B consensus but crazy how quickly its operating leverage can deteriorate despite extraordinary chip demand. Q3 revenue grew 21% to $16.5B, yet non-GAAP gross margin is guided down from 44.1% to 38%. Management attributed $300M of lost Q4 revenue and 200 basis points of margin pressure to the tragic Israel factory accident, explaining only about one-fifth of the revenue shortfall and one-third of the margin decline. The rest reflects weaker PCs, underutilized factories and product mix, with client sales expected to fall over 20% sequentially. Even excluding the accident leaves roughly 40% gross margins, making this much bigger than a temporary shutdown. What makes this especially frustrating is that Data Center and AI revenue increased 78% to $7.3B and generated $3B in operating profit, yet the underlying server figures reveal a more complicated story. Intel's 10-Q attributes server growth to a 54% increase in average selling prices but only 10% higher unit volumes, primarily reflecting richer product mix rather than proven market share gains. Meanwhile, client CPU units fell 14% even as prices increased 9%, suggesting memory inflation is colliding with competitive pressure rather than explaining everything. Management anticipates industry server CPU unit demand growing more than 30% annually through 2027, but Intel 3 manufacturing constraints and EMIB packaging substrate shortages continue limiting shipments. Shifting capacity toward Xeon helps but doesn't solve packaging shortages, potentially giving competitors more opportunities to capture share. The foundry numbers turn this into a question of returns rather than technology. Intel reported $6.7B in foundry revenue, but $6.3B was internal, while only $411M was external and much of that came from a nonrecurring contract, against a $1.85B operating loss. Improving 18A yields and advancing 14A are encouraging, but prospective customers evaluating a process aren't committed high-volume orders. Intel also disclosed additional substrate funding requirements estimated at $5.3B to $6.1B, illustrating how much investment is needed to unlock current demand. Management expects $20B in gross CapEx this year, substantially more in 2027 and likely no positive free cash flow next year. Better yields create shareholder value only if utilization and customer commitments eventually cover depreciation and the cost of expansion. At approximately $65 and a projected 5.44B diluted shares, Intel remains valued around $350B after the selloff. At an illustrative 20x normalized earnings, that requires roughly $17.5B in sustainable annual net income versus about $11B from annualizing Q3 adjusted profits, before factoring in weaker PC guidance. Its $54B of cash and investments against $50B of debt buys time, but the $23B equity raise shows how shareholders are helping finance the turnaround. My bullish scenario requires accelerating server unit shipments without market share losses, improving manufacturing yields and utilization, and recurring external foundry contracts shrinking losses faster than capital spending rises. I'll track those three developments across upcoming earnings because while I remain incredibly bullish on the AI CPU opportunity, I'm considerably less comfortable with Intel's risk/reward until that opportunity translates into sustainable free cash flow per share.

One day left to take your shot at half price. Use code TWO through Oct 31.