Pelagic

Forcing Ukraine to hold elections is a strange take from Trump. Zelenskyy ran as the moderate open to dialogue with Russia in 2019, a stance which has clearly changed but worth highlighting as a starting point. While he'd likely win reelection, the most likely alternative candidate, Zaluzhnyi, is just as hawkish if not more so. Saying Zelenskyy has a 4% approval rating yesterday likely gave him at least a 4% bump from around the 52% he was at in December, but more to my point, what's the end game when it comes to holding elections? Is the thinking someone more likely to cave to Russian demands will win, because if so that seems way off the mark. And as an American what we've gotten in exchange for aid to Ukraine is fairly simple, low gas prices. Ukraine would have almost certainly targeted Russian crude export capacity early in the war and knocked a meaningful portion offline. American aid came with restrictions on what Ukraine could target and this was very much one of the lenses through which the Biden admin prosecuted the war, conditioning aid and capabilities on targets they thought were permissible. That Trumps team doesn't seem to care about this, both in terms of the energy aspect and escalation management, is interesting and likely incredibly confusing to the Ukrainians who've felt micromanaged for the last 3 years. That said, things likely don't change too much going forward as EU aid comes with similar conditions, but the UK and France have both been far less hesitant about strikes in Russia than the Biden team was.

This is to my knowledge the first time Ukraine has targeted a major oil pipeline, certainly the first successful strike on one. Ironically it primarily carries Kazakh oil through Russia for export from Novorossiysk. Crude export capabilities were a major red line under the Biden admin, he even had Harris discuss Ukraine's strikes on refineries last year at the Munich conference, asking them to focus their efforts on the battlefield rather than Russian energy assets. Perhaps a policy shift from Trump, perhaps Ukraine deciding to plot its own course.

Warrior (HCC) is probably the best pureplay US based met coal name. AMR has some thermal exposure but worth looking at as well. BTU is mostly thermal for US operations if you're looking for companies unaffected by tariffs it won't be their met segment. They are shifting the portfolio to focus on seaborne met so if you think US met coal demand>production, they're worth a look too. https://warriormetcoal.com/operations/

NotebookLM is probably your best bet for that purpose. I think there might be a limit of around 40ish files per project, but the workaround (check out the sub-reddit on it) is apparently to merge files into one.

Looking at the proliferation of wire guided drones of late in Ukraine, I think we're seeing electronic warfare capabilities hold pace with advancements in drone tech fairly well. Operators on both sides of the conflict in areas where EW is heavily deployed are basically back to using a TOW missile with far better optics, longer range, and much longer duration. Wire guided drones are also utilized in heavy bomber roles at night and to place anti-tank mines near the front. But with a range of 15km or so max, it's unlikely to be a tech that fits in in deterring an invasion of Taiwan. Furthermore, since a lot of the jammers utilized by both sides are off the shelf products sold on Alibaba, I'd wager China is well aware of the threat FPV style drones pose and has capabilities in place to jam them. For systems that likely will see a major role in a potential Taiwan conflict check out Rapid Dragon and Quicksink, both are a sort of value approach to legacy systems that make them very relevant in the Pacific.

An interesting wrinkle in the Trump aid to Ukraine saga. I think more than anything it's an easy way for Trump to justify continuing aid to Ukraine to his base. The appearance of some form of deal for rare earth metals at an indeterminate date post war in exchange for support - a win - win scenario.

I'm kind of surprised Canada is going with the retaliatory tariffs approach too. I get that their politicians have to be seen as doing something and it's the most straightforward option to implement tariffs in return but there are other options. Putting their head down and committing to a national Manhattan Project style buildout of export capacity would likely put them in a much better position long term. This would also be a good use of tariff revenue in the short term for Canada. If a country (Canada) is selling you resources at a discount because they can't overcome domestic issues to facilitate the construction of export capacity, this isn't a situation you want to mess with, but alas here we are. In a saner world Canada could even resurrect the TPP and spearhead a large free trade union sans US.

Trump's own Treasury Secretary seems to have miscalculated his appetite for tariffs. From Bessent's early 2024 letter to investors: https://assets.realclear.com/files/2024/02/2353_keysquare.pdf

This is a thread on someone who built a setup capable of running the full R1 model on local hardware for around $6k. It's quite slow compared to some of the cloud based models since it's using RAM for inference, 768GB for the full model, and ends up performing at around 6-8 tokens per second. Even the distilled model in the video with a single GPU is doing like 50 tokens/second for comparison. There's probably quite a bit of further optimization possible, and the end user experience of the full model vs. some of the smaller distilled models may not be all that much depending on your application. Very early innings with regard to having models running locally but fun to watch progress. There's probably a decent use for all the old 3060s with 12gb of VRAM that were used for crypto mining here too as they're relatively cheap on the used market.

Isn't inference though where the rubber meets the road so to speak when it comes to end users? The majority of users aren't training custom models but rather using existing models for inference to enhance productivity or build things with it. Companies developing AI models are trying to sell them to users as SaaS, get them hooked on the utility of a limited free tier and slowly let it worm its way into their life until they need the pro plan . It just seems to me there's an appeal to "owning" an open source model to use locally vs. paying a subscription fee for life to one of the big AI players. And back to my original point about this allowing organizations hesitant to utilize AI because fear the data they're providing it might end up elsewhere, having it run on local hardware is a solution at a fairly reasonable price.

Maybe I'm missing seeing more discussion on it, but doesn't the ability to run DeepSeek locally have massive implications? For $4k or so you can build a very capable workstation to run it and this would seem to have major applications for users who have so far hesitated to utilize AI primarily for privacy reasons (think doctors offices, researchers, government contracts, etc.). Running DS locally on an offline machine would seem to be a major advantage in productivity for these potential users at a relatively low cost. And for users paying $200/month for ChatGPT pro, the ability to run a comparative model locally with a breakeven less than two years - and even less if ChatGPT pro price increases, replacing that subscription looks appealing.

I suspect the reference to tariffs might be to third party nations like India and China that trade with Russia in violation of sanctions. Direct tariffs on Russia are fairly meaningless. I also wonder how well informed Trump is on the seized Russian assets and utilizing them as a war chest to provide for Ukraine's defense and rebuilding, it's not an issue he's talked about to my knowledge but something that should have bipartisan support to utilize at this point.

Does anyone have a link to a good study or discussion, or happen to know, why Pelosi seems to favor ITM LEAPS vs. long equity? It seems like she usually ends up exercising the calls at some point as well. I get that the easy answer is insider info but a lot of her success comes in megacap tech, which has done extremely well over the last decade, but at what point did she (more likely her husband) decide that most of their allocations would be in the form of ITM LEAPS.

Less than ideal for all involved. I have to imagine there will be a fairly quick response in gas prices, particularly in the Midwest where most Canadian oil ends up. Maybe Canada gets serious about building more export capacity and it ends up being a net benefit to Canadian producers.

This has always been a major question mark for me as to how/why Russia was able to successfully evacuate Kherson given the circumstances at the time. Ukraine had been hitting the bridges across the Dnipro nightly with HIMARs and they were also in artillery range as Ukrainian forces pushed south to just north of the city before Russia finally decided to pull out. Ukraine also had real time intel from both drones and locals in Kherson on Russian troop movements so Russia's ability to evacuate across the Dnipro with relatively minimal losses seemed odd. Looks like there's finally an answer. Another questionable decision by the Biden admin.