If there’s one thing the semiconductor industry loves, it’s constantly changing the names of products with an alphabet soup of acronyms for us to juggle. Fun, fun. Those acronyms all fall into neat categories: chips, networking and connectivity, and software. It’s these ideas and AMD’s positioning within them that matter to investors. Not that they’ve memorized what an MI325 HBM3E chip stands for. That’s how we’ll frame this coverage, with an emphasis on data center results.

GPU: Graphics Processing Unit. This is an electronic circuit used to process information and data. The accelerated compute needed for GenAI apps and models pulls from next-gen GPUs. It thinks its “MI” series of GPUs (part of the “Instinct” product family) boasts best-in-class memory and bandwidth, which Nvidia would certainly disagree with. AMD also thinks its 2025 Instinct release will compete with Nvidia’s world-class Blackwell platform.

CPU: Central Processing Unit. This is a different type of electronic circuit that carries out assignments and data processing. CPUs fall in the general compute bucket. General compute CPUs are still optimal for static, step-series and instruction-based tasks. They’re also much cheaper than deploying next-gen GPUs when they can work for the specific use case. AMD’s new AI data center CPUs “extend leadership in performance per watt and dollar.” I

NPU: Neural Processing Unit: Used for AI-enabled personal computers (PCs).

TOPs: Tera Operations Per Second. This measures NPU performance, with more TOPs being better. AMD’s new Ryzen AI PC has 20% more TOPs than Microsoft’s best unit. To AMD, TOPs superiority is imperative for running Copilots and GenAI apps on PCs with optimal latency, hallucination rates and performance. It’s how the firm claims to be a “leader in AI inference on the PC side.”

Q3 revenue guidance beat by 1.5%, while its 53.5% GPM guide missed 54.0% estimates. Its EBIT guidance missed by 4.5%. It also raised its data center GPU annual revenue guidance from at least $4 billion to at least $4.5 billion.

AMD trades for 39x 2024 earnings. Earnings are expected to grow by 27% Y/Y this year and by 60% Y/Y next year. Here’s how a 39x earnings multiple compares to its historical norms:

AMD credited its successful quarter to an acceleration in AI business revenue (seen in the data center category). Its traction within accelerated compute (Ryzen and MI processors) as well as networking technology is building.

In data center, GPUs, CPUs and networking are the three buckets to focus on. Its newest Instinct accelerators will be available in Q4 2024, with “leading memory and compute performance.” Its 2025 iteration of the Instinct GPUs will pull from a new architecture (called CDNA 4) to 35x the inference performance vs. its predecessor.

At this year’s Computex event, hyperscalers like Azure debuted more cloud instances using AMD Instinct accelerators; this means more of Azure’s hosted infrastructure for clients is running on AMD hardware. Microsoft is also using the firm’s MI chips (which combine GPUs and CPUs) for Copilot alongside AMD’s ROCm GPU software. ROCm offers a slew of tools and integrations to make using AMD’s chips across applications more convenient. Nvidia’s NVIDIA Inference Microservices (NIMs) and other software tools are seen as having a large edge in GenAI value proposition, so it’s good to see best-in-class cloud players embracing ROCm.

In PC, its newest Ryzen AI processors boast an industry-leading 50 TOPs for AI processing power. It will help run Windows Copilot+.

In networking, with the help of partners like Broadcom and Cisco, it’s piecing together best-in-class tools with its “Infinity Fabric” connectivity tech to emulate the vertically integrated offering that Nvidia has. The “Ultra Accelerator Link” is a group of companies using AMD’s networking tools to build more open, scalable and reliable GPU connections. It will create an “industry standard for connecting AI accelerators.” More connections mean better efficiency, better bandwidth and better client results. 

Chip companies are always iterating and always improving… that is vital when Nvidia is moving as rapidly as it is. And speaking of moving quickly, AMD has committed to a 12-month pace of new GPU platform introductions to rival Nvidia’s “rhythm.” Its Instinct roadmap will stick to this annual pace. 2-3 years used to be a normal cadence for platform launches. Things just move more quickly today.

While GPUs get all of the GenAI hype, general compute CPUs are still quite relevant. And as GenAI raises over compute capacity needs, CPU demand will indirectly benefit in some areas. Wherever CPUs can still be a viable option, companies can realize material cost savings by deploying the cheaper technology. If they can get away with it without sacrificing product quality, they should. Partially thanks to this and an overall rebound in general compute demand, its newest Zen 5 CPU architecture built right into its newest EPYC CPU processors is thriving. It will soon release the 5th generation of its EPYC processors (called Turin). This is set to “extend AMD’s total cost of ownership (TCO) leadership with shipments already starting this past quarter.

AMD EPYC cloud instances offered from hyperscalers rose 34% Y/Y. Instances simply refer to available servers hosted by a hyperscaler like AWS or Azure and powered by AMD processors. These EPYC instances are also in high demand, as companies like Netflix and Uber are using them to power “customer facing, mission-critical workloads.” Oracle’s “HeatWave” offering, which helps customers with accelerated compute and GenAI transformations, is powered by the 4th gen EPYC processors.

“We saw positive demand signals for general-purpose compute in both our client and server processor businesses.”

CEO Lisa Su

Data center was driven by a “steep ramp” of Instinct GPU shipments, which also drove considerable operating leverage. EBIT margin was 26% vs. 11% Y/Y.

Client revenue growth was thanks to Ryzen processor sales. This also drove fixed cost leverage and EBIT margin expansion for it. EBIT margin was 6% vs. -7% Y/Y. 

In gaming, revenue tanked due to a tough environment for semi-custom chip revenue. This led to EBIT margin falling from 14% to 12%. Struggles here will likely continue for the hyper-cyclical revenue segment.

In Embedded, revenue tanked due to continued inventory resets from customers. These are designed for specific apps with less intensive, specialized compute when compared to general purpose CPUs. This led to EBIT margin falling from 52% to 40%. It is “seeing early signs of order patterns improving and expect this segment to gradually recover during the second half of the year.”

AMD announced Silo AI as its 3rd recent AI-related acquisition last week. It has also invested another $125 million across 12 other AI companies. Silo AI Europe’s largest private AI lab and comes with a deep bench of talent to “extend AMD’s capability to service large enterprise customers looking to optimize AI solutions for AMD hardware.” It sounds like this will be an upgrade to its ROCm software. This should make working with AMD on GPU performance optimization, maximizing connectivity, allocating compute and ensuring consistent up-time of that compute power. It could also help AMD build needed application programming interfaces (APIs) to make it easier to service customers using Nvidia GPUs. Nvidia has done a wonderful job of building large reliance on its software suite. That’s partially due to how well the products work and also due to some vendor lock. It’s a lot easier to use Nvidia software with those chips than any other software. This will ideally help in that area.

We’ve heard from pretty much every hyper-scaler this quarter that customers are pushing back against the absurd pricing power that Nvidia commands. Sure, Nvidia’s tech lead warrants that pricing power for now, but many firms are scrambling to try to emulate Nvidia’s best-in-class tech at a more modest price tag. I don’t think any of them will catch up any time soon, but two additional notes here:

It doesn’t need a single chip to be as powerful as Nvidia’s Blackwell (or 2026 Rubin) project. It needs to provide bundles of chips (for likely less margin) to emulate those next-gen products. And it needs strong networking tech partnerships to make sure the processors work well together. It seems to have those ingredients in place. In an overly simplistic way of thinking, it needs 5 Instinct GPUs to cost less than 1 Blackwell GPU and to do as much as that single GPU can do. That’s how it can compete in the near term in accelerated compute GPUs, and the beginnings of its data center explosion point to it making headway.

This is a phenomenal company. Lisa Su is a fantastic CEO. She supplanted a deeply entrenched Intel across PCs and general compute data centers. Still, semiconductors are cyclical and this GenAI cycle will not last forever. The bear vs. bull debate centers around the “how long” rather than if this sector is no longer cyclical… it is. With that said, commentary from mega caps does point to demand not slowing down near term and this data center result shows you that AMD is capable of taking advantage.

Palantir is a software company that helps customers get the most out of their structured and unstructured data. Like many others, it pulls from years of AI/ML work to automate insight-gleaning. It utilizes complex neural networks to power anomaly detection, trend forecasting and natural language processing too. Overall, it frees clients to conjoin disparate data sources while utilizing its software to uncover ideas that manual analytics and legacy competition cannot derive. It gives customers a birds-eye view of their operations, with detailed suggestions to help optimize products and workflows.

Revenue is neatly split into two buckets – “government” and “commercial.” Government clients predominantly use its Gotham product platform, while commercial clients mainly use its Foundry product platform. With Gotham, Palantir routinely builds custom use cases for individual government clients. Foundry was built to be more malleable, with far more pre-built app integrations available. That diminishes the need to conduct custom builds for every single private enterprise. It still does a lot more custom building than a typical enterprise software firm will.

It has also seamlessly leveraged the commercial platform to cater to industry-specific needs. By-industry large language models (LLMs) are intuitively named “micro-models.” These boast sector-specific use cases with granular, relevant regulatory compliance help. A financial services model from Palantir, for example, may specialize in assessing credit risk or fraud detection.

Palantir Apollo is its software suite, which provides continuous integration and continuous delivery (CI/CD) to automate software package building and deployment. It’s a foundational piece of the firm’s ability to collect, utilize and drive value from broad data ingestion. Apollo ties very closely into Foundry and Gotham as a software enabler for both platforms.

Palantir’s newest and most exciting product is its Artificial Intelligence Platform (AIP). This expedites and fully manages model work and deployment for clients. It allows for open collaboration between software developers, data scientists and project managers to ensure collaborative, communicative and effective work. It directly supports Foundry and Gotham by uplifting and augmenting potential use cases. And it does so in a quite compelling way that can craft use cases on a by-customer or by-sector basis.

Considering the the lack of static, finite and structured end products stemming from AIP, I think it helps to hear about some examples of what clients are doing with it: One customer is using it to turn inbound emails into automated inventory decisions, one is using it to automate healthcare documentation for claims and the Department of Defense (DoD) is using it to shrink app creation time from hours to seconds. As leadership will tell you, AIP isn’t just another dime-per-dozen chatbot. It’s an aggregator of data, tools and services needed to actually build valuable apps and to embrace GenAI. It’s how Lowe’s cut overdue task rate by 75% and how General Mills saves $14 million a year in expenses. AIP is where jumbled data, processes and ideas turn into the operationalized, actionable creation of GenAI products.

Initial go-to-market for AIP was its “bootcamps” where it would host events to provide hands-on support and “get clients from 0 to use case in 5 days. It has more recently begun to build out an external sales team to support this segment’s momentum.

AIP progress is most noticeable in its impressive U.S. Commercial results.

Palantir’s free cash flow (FCF) metric is calculated in an unorthodox way compared to typical methods. That’s why the firm’s FCF margin is actually larger than its GAAP operating cash flow (OCF) margin, despite FCF being OCF minus capital expenditures.

Two quarters ago, management told us that it re-vamped comp structures to become more “aligned” with shareholders. That should hopefully mean this rate of increase slows considerably going forward. That progress was never going to begin this quickly. If we zoom out a year and growth is still this elevated, I think that would be a concern for bulls. Not my expectation… but pay attention to this trend.

Palantir trades for 70× 2024 earnings. EPS is set to grow by 33% Y/Y this year and is expected to rise by 20% Y/Y next year. Here’s how its EBITDA multiple (EPS chart is too lumpy) compares to historical norms:

The majority of the call was understandably spent talking about AIP, tangible use cases, how it has built such rapid traction and where it can go from here.

Starting with how it has built such great traction, AIP is actually solving tangible problems for government and enterprises. Leadership, as it often does, ripped on enterprise software. In their eyes, GenAI has made it “very easy to build a prototype” or a fancy slide deck. Bringing those products “from prototype to production” is far more difficult. Palantir removes the bottlenecks inherent in turning GenAI promises into real value.

It does so thanks to the decade+ of foundational work it spent setting up its foundation, infrastructure and organization for this moment. Its software backbone lets companies freely test digital twins (Ontology) in zero stakes environments to actually understand what works and what doesn’t. Its ontology software developer kit (OSDK) allows developers to tap into its data sources and apps to unleash optimal value from this powerful split testing. Palantir’s roster of key large language models (KLLMs) allows it to provide models seasoned with hyper-relevant data to train these assets based on what’s optimal for a client, rather than a one-size-fits-all approach. In Founder/CEO Alex Karp’s eyes, GenAI products and LLMs won’t create value without this foundational software and data processing talent. Palantir provides all of this, in a fully-managed and fully-customized manner.

“The journey to production is fraught and requires a foundational set of technologies that we have uniquely invested in. We created products needed to harvest economic value from AI. And that's a journey that we were pathfinders on with the U.S. Department of Defense starting in 2018. It’s a journey we continue to lead in across segments.”

CTO Shyam Sankar

This led to a very interesting comp from leadership on the call. Like hyperscalers made migrating to modern and managed data center architecture easier, Palantir is making it far less daunting to go from GenAI “prototype to product.”

It’s building blocks like these that are fostering U.S. commercial customer growth of 83% Y/Y and revenue growth of 70% Y/Y (excluding lumpy, meaningless strategic contracts). Bootcamps continue to showcase the potential of AIP and allow Palantir’s team to work directly with clients on crafting the perfect product for them. It has completed 1,025 of these bootcamps vs. 915 Q/Q, and the customer wins stemming from them just keep coming.

AARP went from prospect to client in 45 days, Fujitsu cut $9 million in annualized OpEx in just 3 months with AIP, multiple seven figure deals were signed just weeks after initial client contact, Tampa General Hospital is using AIP to cut patient length stay by 30%, Kinder Morgan is using it to optimize pipeline integrity and its energy grid usage and Panasonic is using it to make EV battery cell production more efficient. Rapid deal closures also directly combat sales cycle elongation that has been plaguing parts of enterprise software for about two years.

To accelerate pace of use-case creation and lower friction, Palantir debuted a free tier for developers to access the tools needed to build GenAI tools. This matters a lot. Companies can’t build everything on their own. They need developers to create apps and optimize their existing apps too. Palantir provides the traffic and demand to make building on its platform attractive to developers. This tool will make doing so a lot easier.

Warp Speed is a new AIP tool being launched specifically for “American re-industrialization.” The team talked a lot about Tesla and SpaceX and how both companies have vertically integrated much of the manufacturing process (hardware and software). The reason? They couldn’t find anything that worked. Products for the sector were created decades ago strictly for CFOs. Product-level software wasn’t a focus, and that oversight has carried into modern American manufacturing. Warp Speed is set to address this. It’s a “modern industrial operating system (OS)” that equips companies and governments with cutting edge enterprise resource planning (ERP), product lifecycle management (PLM) and a manufacturing execution system (MES). It’s a fully managed way to rapidly allow manufacturers to fix how they build things, with AIP integrated right into it to ensure use cases can be as malleable as needed. Karp is especially excited about this product’s prospects.

Mission Manager combines Apollo, its ontology, software development kit and Rubix (to manage apps) into one package designed for government compliance and use cases. It aggregates apps and data to make sure they’re operating cohesively to ensure products delivered via Mission Manager are done so more expediently. On the battlefield, expedience can make all of the difference in the world.

Two interesting notes here. First, the Chief Digital and AI Office (CDAO) system (part of the Department of Defense) awarded Palantir a $153 million contract to deploy an AI-enabled operating system across the Department of Defense. This could become a $480 million deal over the next 5 years.

Secondly, CDAO announced the Open Data & Applications Government-owned Interoperate Repositories (Open DAGIR) initiative. This is set to modernize how the government ingests and unleashes data by allowing developers to build Mission Manager-powered apps and tools on Palantir’s Maven platform. This deal is worth $33 million for Palantir. Palantir leadership is excited about this development, as it thinks it “allows it to ensure access to government-owned, contractor-operated infrastructure while making it easier for companies to develop apps with this data.

This program is essentially the government more openly collaborating with the private sector and inviting them to help them build what they need. That’s good for Palantir and all of the partners that it has under its full service Impact Level 5 accreditation program called FedRamp. Its First Breakfast product, which expedites time to product eligibility within the Department of Defense, should help a lot here too.

The government segment was partially helped by favorable deal timing, which likely was a source of some (not all) of the outperformance this quarter. Palantir closed 96 $1 million deals and 33 $5 million deals during the quarter. It notably closed 27 $10 million deals vs. just 15 of them Q/Q. U.S government revenue rose 8% Q/Q vs. 8% Q/Q last quarter and 3% Q/Q two quarters ago. International government revenue rose 18% Q/Q vs. -9% Q/Q last quarter due to ongoing efforts in the Ukraine, easier comps and contract timing.

U.S Commercial revenue rose 6% Q/Q vs. 14% Q/Q last quarter and 12% Q/Q two quarters ago. Commercial revenue excluding strategic contracts rose 8% Q/Q. U.S. Commercial customers rose 13% Q/Q vs. 19% Q/Q last quarter. U.S. Commercial average contract value (ACV) rose 19% Q/Q.

Total Commercial revenue excluding strategic contracts rose by 8% Q/Q compared to 3% Q/Q growth with these contracts. International Commercial revenue fell 1% Q/Q due to Europe macro headwinds.

Billings, a key forward-looking demand metric, rose 19% Y/Y. Overall, it closed almost $1 billion in TCV representing 47% Y/Y growth and net revenue retention (NRR) rose to 114% vs. 111% Q/Q. This doesn’t reflect the fantastic momentum from new AIP customers over the last 12 months. Palantir’s rule of 40 score rose to 64 vs. 57 Q/Q. That’s excellent.

This was simply phenomenal. Bears could pick on government deal timing propping up revenue or dilution, which is fair, but I found this print to be overwhelmingly positive.

It shows you exactly how powerful of a monetization engine GenAI can be for enterprise software firms that actually build good products. It brings real credibility to leadership’s opinion that it’s well ahead of the competitive pack here. Palantir has leapt into the elite company of Azure in terms of its ability to drive financial success from this technological wave before most others can. AIP is rocking, margins are exploding, the product pipeline is humming and Palantir is killing it. Alex Karp’s lengthy monologues and bold predictions have been routinely picked on by investors for years. These results and the stock price should take care of that.

Hims beat revenue estimates by 4.3% and beat its guidance by 7.1%. Its 2-year revenue compounded annual growth rate (CAGR) actually accelerated Q/Q from 66% to 67%. That is rare for this earnings season in the best of ways.

Among GAAP cost buckets, most of its operating leverage came from sales & marketing falling from 51% of sales to 46% Y/Y. This was thanks to “significant efficiencies” related to more cross-selling and strong top-of-funnel trends. G&A moved from 15% of revenue to 13% Y/Y, R&D was flat at 6% of revenue and operations and support moved from 14% of revenue to 13% of revenue. Gross margin fell modestly Y/Y due to new product investments.

Q3 was also well ahead of estimates across the board.

Hims trades for 36x 2024 EPS estimates. EPS is expected to grow by 137% Y/Y this year and by 49% Y/Y next year. Estimates are absolutely going to materially rise after this report.  It has not been profitable for very long, so here’s how its current gross profit multiple compares to its historical norms.

“We believe healthcare is undergoing a transformational period, as consumers are starting to expect many of the same benefits that technology has unlocked across other areas of their lives. These include transparent pricing, convenience, speed, and affordable solutions.”

Shareholder Letter

At the end of the day, Hims sells medications online. When considering this, one may rightfully wonder: “Can’t Amazon just do this instead of them.” As Hims gets larger and larger, I see that likelihood growing over time. But? Hims is beginning to establish unique competencies that could help it defend market share over time. All of which involve medication personalization. Success here can be seen in 164% Y/Y growth in personalized subscribers to reach 785,000, or 42% of total vs. 22.8% of total Y/Y. It’s one thing to be discrete, great at marketing and small enough to not invite unwanted attention. It’s another to provide valuable products that folks can’t get elsewhere.

It’s (shockingly) leaning on 7 years of first part data and customer learning to drive more personalization in several ways. These include targeting multiple conditions with one product, balancing dosing and formulas to minimize side effects and offering flexible form-factors for people who struggle with pills and injections. It can even insert vitamins and minerals into a single prescription to make it healthier for an end user. All of this work led to personalized offerings in sexual health rising 70% Y/Y and also 85% of its new Dermatology subscribers opting into a personalized solution. These personalized products not only insulate Hims from competition, but raise retention and lifetime value too. It will soon launch a new option here that combines erectile dysfunction and hair loss for men. It will start at $49 per month, and management expects it to have a “notable impact on customer acquisition and retention.”

MedMatch is the firm’s AI platform to power this data-driven personalized. It sees these models simply becoming more powerful over time (I would hope so) to unlock more variables to perfect and personalize.

Leadership’s confidence in all 5 of its current categories, reaching $100 million in annualized revenue by 2025 rose sequentially. It thinks GLP1 injections (much more later) will be “incremental” to that trajectory. Growth for its most mature products isn’t really slowing either. Its ex-GLP 1 revenue rose 46% Y/Y to eclipse $300 million. Every product category that it plays in is growing nicely.

“We see a clear path toward serving tens of millions of customers as we scale this increasingly powerful and efficient model."

CFO Yemi Okupe

Hims started with oral medications within weight loss, not GLP1 injections. And? The oral segment alone was already at a $100 million revenue run rate ahead of schedule. It also reached 100,000 subscribers here in just 7 months, vs. 18-24 months on average for a new product. Still, GLP1 injections are the most exciting growth lever here. Like all other areas Hims enters, obesity is plagued by lack of information, lack of coverage and poor access. It’s also an emotionally-charged issue, is chronic and has a large addressable market. That’s all perfect for Hims. Hims is looking to fix these issues with its typically convenient and subtle deliver, but also with a large cost advantage. For context, it will charge as low as $70/month for these drugs vs. $1,000/month on average (or 25% of an average patient’s income). Crazy costs are part of the reason why 58% of GLP1 patients stop using the medications within 12 weeks. Proper dosing via more personalization, more comorbidity consideration and better provider communication, it thinks, will help here.

Hims acquired an outsourcing facility to service strong demand in weight loss and other areas down the road. This builds on its 355,000 square feet of affiliate capacity (including pharmacy partners) and represents its “first step to verticalizing this piece of the support chain.” That implies there will be more steps here and that this is a priority. Why vertical integration? More ability to drive deeper customization, control service quality and cut cost. It will also keep investing in robotics to automate more parts of its fulfillment footprint over the next 3 years.

This was a fantastic quarter. I worry that competition or regulators (pushed by deep-pocketed companies like Eli Lilly) will eventually have a material impact on this business. Today, that is not happening. I’ve never found this business model compelling, and here it is today absolutely thriving… This is impressive in every sense of the imagination.

The growth engine is wonderfully rapid, as is margin expansion and continued marketing efficiency gains. This firm has a seemingly endless supply of productive external marketing dollars to spend and should find even stronger returns as it broadens the product suite. Personalization is its secret weapon for combatting new entrants over time, and trends here are positive. So far so great for this team and this just 7-year-old company. What a story.