
Swollard
412 posts

Swollard
@SwollardCapital
Investor and Entrepreneur| Family Portfolio Manager| 7+ years investing+options| Not financial advice


Gavin Baker @GavinSBaker says open source is "awesome" for AI infrastructure names because it shifts the economic value from the frontier labs to the infrastructure, and it is already 80% plus of all tokens processed. "A misconception that a lot of people have is that open source models are somehow bad for AI. They're awesome for the AI infrastructure providers. They just shift economic value from the margins of the frontier labs to the infrastructure." "But I do think there's still a role for these frontier models. And it may be true, to date, frontier tokens are capturing 90% of the economic value, and open source tokens are probably 80% plus of tokens processed, and those ratios may be here to stay." _______ Follow @firesidealpha for more highlights of the best business and technology conversations.

Deep|LLM: Kimi K3's KV Cache Is Smaller - Why That May Actually Be Positive for DRAM/NAND K3 is still a large-memory model: Kimi K3 has 2.8T total parameters, activates 16 of 896 experts per token, and Moonshot recommends deployment on high-bandwidth supernodes with at least 64 accelerators. Its smaller KV cache does not mean lower overall memory and networking demand; K3 still needs HBM-scale capacity, scale-up networking, and GPUs. KV offload becomes practical only after compression: Large uncompressed KV caches are hard to place on the active decode path because PCIe, NIC, and SSD latency can erase the benefit versus recomputing prefill. KV compression reduces the data that must move across DRAM, NAND, and the network, turning offload from a theoretical capacity extension into a deployable architecture. DeepSeek V4 shows the threshold effect: DeepSeek V4-Pro cuts KV cache to 10% of DeepSeek V3.2 at 1M context, yet channel checks suggest NAND usage has increased in real deployments. The reason is that smaller KV enables higher offload ratios, more retained sessions, longer prefixes, and better cache-hit economics. Implication for DRAM/NAND: The key trend is not KV cache disappearing, but KV cache becoming a tiered data asset across HBM, DRAM, NAND, and networking. DRAM becomes more important as a warm-cache and staging layer, while NAND becomes a large-scale pool for historical sessions, shared prefixes, and inactive KV. Detailed Report fundaai.substack.com/p/deepllm-kimi…

Data centers barely use any water, barely use any land, and they lower electric bills.

ANTHROPIC WILL BE AN AMD CUSTOMER, ACCORDING TO THE PUBLIC GITHUB OF AMD’S SENIOR DIRECTOR OF AI 🚨🚨 We explain the GitHub code and nuances below👇️ 1/4🧵


MERITZ SECURITIES: ACCORDING TO CHANNEL CHECKS, MIDDLE EASTERN SOVEREIGN AI INVESTORS, INCLUDING THOSE IN SAUDI ARABIA, HAVE RECENTLY BEGUN DISCUSSING MID- TO LONG-TERM MEMORY PROCUREMENT PLANS WITH KOREAN MEMORY MANUFACTURERS. - Amid this demand growth, upward pressure has begun to emerge in the server DRAM spot market. - Prices are rising particularly sharply for high-end products, such as those with 6,400Mbps bus speeds. - This suggests that intensifying investment competition among CSPs and frontier-model developers is increasingly focused on products designed to maximize performance, amid worsening supply shortages. - Spot prices for 64GB DDR5 server DRAM products have risen steeply since mid-July. - Recent prices have climbed to $3,100–$3,400, approximately 146% above the end-June contract price of around $1,380. - ACCORDINGLY, SERVER DRAM CONTRACT PRICES IN 3Q26 COULD RISE BY MORE THAN THE MARKET’S CURRENT EXPECTATION OF APPROXIMATELY 15% QOQ DURING THE QUARTER. - In particular, suppliers that adopted more customer-friendly and flexible pricing in 2Q26 are likely to see especially sharp price increases in 3Q26 and 4Q26.

polymath book recs + seeking new recs










