Syz the moment

🚨 In the wake of DeepSeek turning the entire industry on its head — and wiping nearly $600 billion off of the market cap of Nvidia in a single day — one new phrase has become table stakes for anyone wading into the DeepSeek discourse: Jevons Paradox, with traffic to its associated Wikipedia page soaring this week. 👉 Per that very Wikipedia page: “...the Jevons paradox occurs when technological advancements make a resource more efficient to use (thereby reducing the amount needed for a single application), however, as the cost of using the resource drops, overall demand increases causing total resource consumption to rise.” 👉 The original example posited by Mr. William Stanley Jevons, summarized nicely by Axios, was coal. Progress in steam engines, which enabled them to use less coal, didn’t lead to a drop in coal demand — it led to a huge rise. Though a bit of an oversimplification, that is essentially the crux of the current debate in AI: DeepSeek reportedly achieved something for a lot less money and resources than US competitors like OpenAI and Meta used. That could be interpreted in two ways: • We will therefore need fewer high-tech chips like the ones Nvidia makes, and fewer energy plants to power them (which is why power and datacenter stocks got hammered this week); • Or, and this is where the Jevons Paradox comes in, WE WILL WANT EVEN MORE 💪 The market seemed to follow the first school of thought on Monday 🐻 , but came around to the second by Tuesday 🐮 , with chip analysts and tech heavyweights, most notably Microsoft’s CEO Satya Nadella, citing the paradox as proof that AI use will “skyrocket.” 🚀 🚀 🚀 Source: Chartr

As the cost of AI comes down, what are the sectors that benefit from cheap intelligence? 1. Cybersecurity (e.g $CRWD) 2. Data Storage/Analytics (e.g $NOW) 3. Robotics (e.g $AMZN) 4. AI Agents (e.g $MSFT $CRM) 5. Advertising (e.g $META, $GOOGL) 6. Ecosystems (e.g $AAPL) NB: These are not investment recommendations Source: Lin@Speculator_io

With the rise of DeepSeek & other small models, questions are the following: 1) Will these giants maintain their spending forecasts? 2) How they justify it after DeepSeek release? 3) Will the perceived value move up the chain towards applications? Source: Deutsche Bank thru Ali Dhanji @DhanjiatRJ on X

“Our view is what happened with DeepSeek is actually bullish because it advances the move to artificial super-intelligence” - Bloomberg Source: Evan on X

Source: SemiAnalysis via @FT

Source: Bloomberg, Adam Kobeissi

Did they short nvidia before announcing the world - through a paper authored by their lab - that the DeepSeek-R1 model outperforms cutting-edge models such as OpenAI’s o1 and Meta’s Llama AI models across multiple benchmarks?

>> Studies machine vision at Zhejiang University >> At 30 in 2015, launches High-Flyer quant hedge fund >> Makes a fortune (now $8B AUM) >> Wants to build “human” level AI as side hustle and pitches partners but they initially sceptical >> Buys 10,000 H800 chips in 2021 and brings over his top hedge fund employees (all have tons of experience squeezing juice out of Nvidia GPUs for the fund) >> Launched DeepSeek in 2023 and hires dozens of PhDs from top Chinese universities (Peking, Tsinghua and Beihang) >> Pays top top top salary for tech talent only matched by Bytedance in China…wants DeepSeek to be leading “local” company >> US export restrictions force DeepSeek team to get creative and they do, finding new training methods to make LLM models (V3, r1) competitive with OpenAI, Anthropic, Gemini, Grok, LLama etc at ~1/20th the cost >> Training costs not exactly apples-to-apples but novel methods and clear improvements in efficiency (also questions around copying other models, larger H-100 clusters they maybe can’t talk about and/or CCP support) >> Open sources and publishes methods (r1 reasoning paper has 200+ authors) >> DeepSeek just hit top of App Store *** FT: https://lnkd.in/e96ffxmU Source: Trung Phan @TrungTPhan on X, FT