The Perfect Storm: Palantir and the Future of Software
Вставка
- Опубліковано 27 вер 2024
- This episode breaks down industry trends, Palantir execution risk, and what I see as a potential game-changing software architecture using the Game Plumbing Framework, Foundry, and Apollo.
Join the discussion: / discord
Thank you so much for these videos! As a non technical investor my DD is done mainly by reviewing the finalcials and whatever customer testimonials I can find. These videos bring in a whole new level of understanding and really top off the diligence process. You're the man Code!
Thanks so much! Social investing brings all the insights together IMO.
We are in it together! Thank you Palantir family
We need Codestrap to interview Alex Karp. Make it happen internet.
An interview with Akshay would also be nice.
Execution risk is sobering. Excellent post as usual, thanks.
Thanks!
I look forward to hearing the words “welcome back to the channel everyone” every single week
💯
Thanks for your dedication to present to the public the nitty gritty details of the tech and also to keep a balanced view of the investment thesis by outlining the risks. This is very valuable.
Just finished watching the video while at work. Very detailed and really helps non technical and non programmers understand the importance of Apollo and Foundry. I just hope they execute well.
Thank you for posting this sobering video. Good to hear both sides in a stock I am asymmetrically invested in.
Thank you!
The video is not sobering. I think it’s more positive than negative. Every company on the planet has competition. As long as PLTR takes majority of the market share, that’s all that matters.
I’ll watch this at work. Trying to catch a few 💤 before I go in for my night shift. Just posting this comment to help move the algorithm and to thank you for the constant content.
Thanks so much for your amazingly explanatory videos!
Can’t wait for this. Will save it for my morning drive Monday
Nice mirror image view of Test Baker. Battleship Arkansas is rising up on the wrong side of the pillar of water.
The way sky compute works kind of reminds me of pied piper lol. But this is great content - keep it up, man!
Awesome CodeStrap. Keep it coming, appreciate the effort and explanation. To the moon 😁👍
Loving the content. Keep it up man. 👍👍👍
If Palantir isn't a major player in the industry and doesn't capture significant market share overtime, it would be a tragedy. So much talent and innovation, but the execution is TBD.
Honestly if in the next 5 years foundry is not widely adopted Palantir will have no one to blame but themselves.
I agree but I wouldn’t bet against Peter Thiel and Alex Karp. Also, remember they have worked in government for over a decade. They have a level of real world experience that their competitors don’t have
That ship with that music is perfect intro
Thank you so much Code!!! Big appreciate for everything you did.
Palantir has a product "Data Mesh" that only release in Japan. Could you help us understand more about it? Is it a small version of Foundry?
I'll check it out!
Awesome video strap. Worthy
Thank you!
Yea buddy! Party On with some Michter's Rye!✌️😃✌️😃✌️
👍
Can't smoke it drink it 🤪🤪✅
@@Soto24 pro tip skip a drink and buy a pltr share(;
@@Soto24 Yea buddy🤣🤣🤣🤣✌️✌️✌️✌️
@@cameronvincent For sure! Now we are talking 'ol pal!👍👍👍👍✌️✌️✌️✌️✌️✌️
I'm glad palentir is listening 👍 I hope they are...
I hope some of these questions you raise are answered at our next meeting
Excellent deep dive!
Hope you don't mind some stupid questions though:
The one point I did not get is the precise nature of the "execution risk" regarding Apollo(?) compared to VMware Cloud Foundation.
Is your criticism that Apollo is currently not accessible for try-before-buy?
Or is your point that Palantir should be a cloud hosting provider ("Palantir hasn't created their own cloud yet") ?
Why do you think Palantir needs to "create their own cloud"?
(Why) Isn't it enough if Apollo supports Azure, GCP, AWS + on premises ?
As far as I'm aware they are offering Foundry+Gotham as cloud offerings - presumably a SaaS managed/operated by Palantir (using Apollo as the tool) and running on one of the big 3 platforms. Not good enough?
Final Q: Is that the only major technical execution risk that you see (i.e. apart from scaling customer acquisition)?
Awesome questions! The criticism is both of those things. Being your own cloud hosting provider doesn't necessitate you own the infrastructure, it means you can own the relationship first-party (sign up with credit card on my website) and allow the customer to choose their infrastructure (abstract away all the cloud complexity ie Sky compute). VMWare is a good example because you have the option to both sign up for their service on their site or integrate the service using your cloud hosting provider. PLTR only offer the latter. In. regards to Apollo specifically there are actual advantages if you own the underlying infrastructure. A good example of this is AWS Lambda and the changes it underwent to solve cold start issues by creating Firecracker. Because AWS owns the underlying infrastructure they were able to solve things at the network layer as opposed to just the virtual machine layer. Apollo could be superior in many ways if it too could offer an edge network to augment existing infrastructure with features that simply are not possible anywhere else. This is widely known in the industry and I would expect engineers at PLTR actively debate such things (I do not know if that's the case but I would assume so).
Regarding the current SaaS offerings on all major cloud providers, it's essentially invite only with upfront sales to access to the platform. There's also not much in terms of training which does require some sort of freemium access. I think this is what leads analysts to label them a consulting company (which is not the case in terms of their products).
Lastly no, those are not the only technical risks. I'm speaking for myself here, speculatively. I have no information from PLTR on the subject. There are potentially much larger problems that require total platform rewrite, or at least a major refactor. I've done several videos on this topic including "Why PLTR wants a very big pool" and even my very first video covering PLTR over a year ago where I could see they use a silo model for SaaS. Scaling customer acquisition at a software company is the primary business metric constrained by tech debt, so it's a big risk.
Hey Codestrap, do you think you could explain why the architecture you’re proposing would lower the compute bill?
If you built a synchronous architecture, you’d be hanging at multiple layers for an action to finish, whereas with this model, the only compute being used would be on the Foundry “server”. Is that the correct way of looking at it?
Not really. Imagine you have a very intelligent system that moves the server objects (the things that actually stream data to the client and manage mutations) around a global compute infrastructure to land the system as close to the user as possible, including their device. This same system can handle things like noisy neighbor problems routing users who consume lots and little amounts of data to shared server object instances. This type of architecture is purpose-built for edge-enabled applications that scale up or down over a global computing network. This type of architecture has the benefit of using the cheapest computing possible for the most scalable parts of the system. You can think of this as an application serving layer for Foundry. Foundry sits in the background and can leverage eventual consistency using the data streams as a throttle (mutations can persist in streams for hours or even days in extreme cases). You are trying to frame within a synchronous vs. asynchronous paradigm and that's not right. You need to think in terms of eventual consistency and eventing. This architecture isn't unique to Foundry, but very, very hard without Foundry.
@@codestrap8031 gotcha. So if another company did this with their own architecture stack, why would it be cheaper to do so on Foundry
@@Foremangrill Because eventual consistency problems like this is hard. Ontology-based systems are hard. Building data integration tools is hard. Buiding low code/no-code application tools are hard. Like 10 years of work assuming good execution hard. If another company wants to try (and this had been the level of stupidity to date) go for it, see you in three years when the wheels fall off (and if they succeed, maybe they should get out of their existing business and compete with PLTR). That's a large reason why I picked 2025 as the inflection point for mass adoption. That's just long enough for everyone to admit they failed.
@@codestrap8031 cool thank you so much for going into detail on this. I would love to hear more about the eventual consistency architecture if you’d be able to do a video just specifically focusing on it and showing the code along with it.
One other question I do have after reading a bit more on eventual consistency is what happens in use cases where data needs to be up to date? Maybe like with stock trading or something? Are these systems just not able to benefit as much from Foundry?
nice on the intro
Nice video.
Thanks!
Thank you Code !
Guys LIKE the video. Likes are free.
Thanks as always. Your tone sounds quite doubtful however…makes me nervous
Hahaha. I wanted to reset the tone for my viewers. Nothing I said here is new really. If you go back to my very first video I call out the risks, and those are largely unchanged. I even did a red flags video that also call the strategy out. My eyes are wide open to the risks associated with PLTR, and they always have been. But many people who watch my channel get overly confident IMO. I wanted to kind of wake people up a bit and make them realize it's not a bed of roses. Nothing is.
@@codestrap8031 Heard 👍
Every company has risks and Palantir is no different. How Palantir do with these risks will require investors to really question them and ask them during conference calls instead of the usual SBC question.
Love it! Thanks!
WOOOOOOOOOOOOOOOT
For my feeble gamer mind.. are PLTR products like an extreme version of WeakAuras (World of Warctaft Addon ) but for real life?
Servers and storage is beeing putt in places most ppl can’t even imagine. We are talking enterprise grade stuff around you every day. Edge is on the rise baby. By 2030 atleast 75% of data will be created and processed outside of data centers and public clouds
Bag holders only can see what they want to see, not a chart.
The stock price is not the company and the company is not the stock price.
so why are you here then lmao
I really hope Palantir sees this video. As an Investor I'm very scared of the execution risk. I can't believe that such a superior product is not getting more traction
I really hope PLTR watch your videos or you have direct coms back to someone that listens. Do you think there are any good acquisition opportunities for PLTR out of the companies you mentioned ?
Stackblitz for sure. I did let PLTR know they should consider them for an M&A opportunity for many reasons. Their web based IDE would be amazing in Foundry. Also the have web containers which Apollo could integrate as a deployment target. I don't know if they are thinking about it or not.
@@codestrap8031
CodeStrap
Strategic Advisor
Palantir Technologies
@@deemahdee 💯
@@codestrap8031 Regardless keep giving them your opinion. You truly know what you're talking about and see where the future is going.
Is one of the reasons for the lack of ubiquitous availability of the software due to the fact that anyone (i.e. foreign adversaries, irrational actors) could then use the platform and Palantir wouldn't have control? If so, then what would be the way around this?
I dont think so, at least not for thier SaaS sales since those workloads run in PLTRs infrastructure. There might be InfoSec concerns though. Maybe they are afraid it would increase the attack surface of Gotham.
CodeStrap, the legend we needed in the pltr community
Thanks so much Codestrap love the video. 23:21 What are you looking for in terms of the approx number of customer count for the end of 2022 to be an either ok number or great number. Thanks
Not too much this year. This year id prefer they finish the foundational work to enable customer growth in 2023. 1000 customers is the target by 2025. I'd like to see 10k more in the next two years and 100k by 2030.
awesome thanks for your insight I value your thoughts for my consensus
Path is clear for Palantir....see you must young Jedi's..... Thanks for the info
Good video, I appreciate all the effort you put in. Thanks.
Thanks again CodeStrap, you're videos are amazing!
Thank you!
Comment for algorithm
PLTR has an image problem. They come across as mysterious and clandestine. Need a PR makeover.
Great video as always! With compute moving to the browser when you were explaining webcontainers, do you think a consumer facing Palantir product could be a web browser?
In my mind if Palantir is already building for this future wouldn’t they be best suited to build a web browser that best enables this future? Could also be strong for privacy, appealing to everyday consumers who may be getting more concerned about their web privacy. Maybe even Palantir logins for apps, like a Google or Apple login. Not sure if any of the above would make sense. BTW, I’m a marketer, I don’t know a thing about building software. Let me know if I’m way off here.
They won't be making web browsers IMO. They don't need to to take advantage of web assemble which is what powers server gateways running in the browser. Its an open standard developed by a working group. They could join the WASM working group though. Where PLTRs products fit in is the software management side at the edge using Apollo and moving at the speed of the data stream using Foundry. I wouldn't rule out consumer products related to KYC and security applications though. Those could all plug into existing browsers.
@@codestrap8031 awesome. I have lots to learn. Thanks for the quick super detailed response. Looking forward to future content.
Would Greg DeArment be able to answer your question about execution risk? Would he even be allowed to answer? Or does it simply come down to Karp?
I don't think anyone at PLTR would be able to share that information. Maybe Karp could share the details, but not sure it matters why they've executed the way they have to date. It's more important that we know what's on the roadmap now and when these things will be "fixed" (from my perspective anyway). I could also be wrong and the way they've executed to date is better than say C3 or VMware.
Palantir will keep delivering Alpha and companies will either invest and reap rewards or suffer inflationary pressure from their inefficiencies. Their having missed the boat on datalake services has two possible explanations 1- They were always planning on a far more sophisticated dominant position with enterprise ontology and not simple columar databases and integrated ML apps and 2- They were still very busy developing infrastructure for the US military etc. and hadnt yet had the opportunity to expand into the civilian markets. If you sidetrack from the masterplan to follow short term profits it will be very difficult for you to build something competitive enough that dislodges the current big tech powerhouses.
For instance could we claim that Snowflake`s moat is anywhere near as big as Apollo`s?
All those are plausible, but still a total failure to execute. With the benefit of hindsight we can see the market opportunity for those "thin products" is substantially larger than anything they've delivered on to date. They weren't short term gains by any stretch. That's doesn't mean the master plan they are working on isn't potentially bigger out past 2022. It does mean that 100% for sure billions upon billions of market cap and revenue was left on the table for their competitors. Executive leadership doesn't get a pass for that no matter the reason, at least that's been my experience dealing with investors. You have to capitalize on your opportunities, and if you fail to see fully what those are you are out. It really took a lot for Karp to own up to that in the shareholders call. I respect that. But I expect nothing but perfection from here on out.
@@codestrap8031 You are right at the minimum it would have meant a lot of cash flow they could be spending on sales and marketing right now for instance. And if they had already solidified a client base with those initial services imagine how quickly they could be rolling out foundary and apollo right now.. The whole organization knows how important this next year is so ill be looking forward to what they achieve but at the same time if they start to underperform it will be a sign to diversify more and pick up some other ML tailwinds.
Thank you, Code
It was my pleasure. Glad you enjoyed the video.
Thank you!
Thanks,