Michal Hlavacka @Michalhla: "𝐏𝐨𝐫𝐭𝐟𝐨𝐥𝐢𝐨 𝐮𝐩𝐝𝐚𝐭𝐞 - 𝐰𝐞𝐞𝐤 𝐞𝐧𝐝𝐢𝐧𝐠..."

Czech Republic

𝐏𝐨𝐫𝐭𝐟𝐨𝐥𝐢𝐨 𝐮𝐩𝐝𝐚𝐭𝐞 - 𝐰𝐞𝐞𝐤 𝐞𝐧𝐝𝐢𝐧𝐠 𝐉𝐮𝐥𝐲 𝟓 Past week returns: 🟢 This Portfolio: 1.44% 🟢 $SPY (State Street SPDR S&P 500 ETF) 1.76% 12 month returns: 🟢 This Portfolio: 34.75% 🟢 S&P 500: 19.17% 2 year returns (since Jan 2024): 🟢 This Portfolio: 207.3% 🟢 S&P 500: 56.68% Since March 2020 (start of the portfolio) 🟢 This Portfolio:+1042.02% 🟢 S&P 500: 154.24% Check out more statistics about the portfolio here: bullaware.com/etoro/Michalhla 𝐂𝐡𝐚𝐧𝐠𝐞𝐬 𝐢𝐧 𝐭𝐡𝐞 𝐩𝐨𝐫𝐭𝐟𝐨𝐥𝐢𝐨 No changes 𝐍𝐞𝐰𝐬 $RKLB (Rocket Lab Corp) is acquiring $IRDM (Iridium Communications Inc) - a communication satellite network operator. The company will get a head start on having its own constellation, which was always their long-term goal. It will get a working and profitable constellation with fairly new satellites and valuable spectrum for radio communication that works in harsh weather conditions. The acquisition will also double the revenue and make $RKLB as a whole profitable despite the capex-intensive Neutron development. I believe this will speed up $RKLB’s space business flywheel greatly. Neutron will have internal payloads to launch for the Iridium constellation and I have no doubts the constellation will be upgraded and given further capabilities to drive more revenue. $META (Meta Platforms Inc) announced they will sell excess AI compute similarly to $SPCX (Space Exploration Technologies Corp) Investors liked the potential for additional revenue for the company while the market generally interpreted it as a possible signal that the AI buildout is overdone. Neoclouds like $CRWV (CoreWeave Inc) went down as a result. My view is that it is an opportunistic step to boost the share price, following similar moves by SpaceX that have worked well to support its IPO valuation. In Meta’s case, I would not be surprised if the company had overbuilt the AI datacenters, as it does not have a compelling AI product for external customers and won't be able to consume all of it just for better ad and content serving. $NVDA (NVIDIA Corporation) announced a partnership with $PLTR (Palantir Technologies Inc.) which will use Nvidia’s open model Nemotron in its new reference architecture called “Sovereign AI Operating System”. While it might seem like minor news, this might be an inflection point and $NVDA might be on a trajectory to directly compete with leading AI firms to which it used to be a chip supplier. This seems to be a part of a broader trend when many prefer not to use the Leading AI providers, as they have repeatedly demonstrated they use data of their users to build a competing service in-house. In this particular case of use by Palantir, it is also linked to the Anthropic clash with the Department of War, where Anthropic denied all lawful use of its models and was later designated as a supply chain risk by the administration. The clearest example of the first problem is Anthropic releasing its coding agent Claude Code after it has been providing API service to the prior most successful coding agent by Cursor. Open models like Nemotron can be customized and further post-trained using specific data and know-how and held by the user (in this case, Palantir) in order to make them on par with the leading LLMs without facing the risks outlined above. These solutions will be also run on premise - using in house hardware to avoid sending data to another party. $NVDA is definitely not happy to see Anthropic and OpenAI being very promiscuous with use of compute and chips outside of the Nvidia ecosystem like TPUs from $GOOG (Alphabet), Trainium from $AMZN (Amazon.com Inc) and GPUs by $AMD (Advanced Micro Devices Inc). Moreover, both have their in-house chip design initiatives. Nvidia’s Nemotron open model is tailored to Nvidia’s chips, and pushing for its broader use also instances where the leading models were previously used should help to ensure market for $Nvidia chips against the intensifying competition. It is hard to model what will be Nvidia’s next steps but one of the scenarios is that it trains and releases its own closed leading model if Anthropic and OpenAI continue to leave the Nvidia chip ecosystem. Being able to sell AI services running on their own hardware would create a competitive advantage. Nevertheless, to date there are no clear signs the company wants to build the data centers and infrastructure needed to do that. $ASML (ASML Holding NV) raised revenue guidance slightly due to higher demand. Disclaimer: Copy Trading is not investment advice | Capital at risk | Past performance does not guarantee future results

Not investment advice. The author may have financial interests in the mentioned instruments.