How close are current LLMs in performance?

In this episode, Doug sits down with Gil Luria, head of technology research at D.A. Davidson, for a wide-ranging and candid discussion on the current state of the AI arms race.Key Takeways:OpenAI has over-promised on the $1.4T of commitments to ORCL, CRWV, etc… and likely has to renegotiate and reduce the scope of its ambitions across the AI value chain to focus on their core like ChatGPT. There are big competitors in the LLM race, such as GOOG and META, so OpenAI will need deep pockets that are willing to bet on Altman to keep the story rolling. Raising capital in the private markets is the near-term solution to solve the funding gap…and all is ok at OpenAI as long as the funding keeps coming and the growth keeps beating expectations….not a lot of margin of error here. If it’s a game of who can lose money the longest, I would not want to bet against GOOG or META’’s cash flow machines. META could find discipline and reduce capex, don;t hold your breath as Zuck has a big ego and a big balance sheet. He’s sending a signal to the market that he is going after AGI and has the balance sheet to do it. Stock would be better off if META split into an infrastructure company and an asset light social media company. The stock price driver in the near-term is based on capex and Zuck’s desire to win the LLM race and AGI. Ugh.GOOG is now the consensus winner, low-cost AI provider and now with a SOTP story. Regulatory breakup behind them and now commercializing TPU’s and marginally ahead in the LLM race. If OpenAI starts to monetize its traffic or new position in the TOF, then it could pose a risk to the consensus long GOOG…No signs of that yet. Will check CarbonArc for the latest digital advertising alternative data. The OpenAI contracts with ORCL and CRWV are “known” to be be unattainable and if they are renegotiated, it could potentially be a positive catalyst for those related stocks that have already discounted a negative outcome depending on the new terms as the negative catalyst is past (potential short covering) or the overhang is removed. Disclosure: Not Investment advice. Host andor affiliates are long GOOG and META and short NVDA.📩 Subscribe for more deep-dive conversations and variant views: https://pitchthepm.beehiiv.com/subscribeDoug Garber on LinkedIn:https://www.linkedin.com/in/doug-garber-42aa508Episode partner:💡 Powered by AlphaSense. For complimentary access to their GenAI and expert call library, visit: https://www.alpha-sense.com/Pitch/