DeepSeek's DSpark Just Made Nvidia's Most Important New Bet Harder to Close
View original at nasdaq.comDeepSeek's DSpark Just Made Nvidia's Most Important New Bet Harder to Close Key Points Nvidia's GPU business is not in question. What is in question is the incremental hardware layer Nvidia is now trying to sell on top of it…
What we drew from this source
The claims Via News extracted from this document. We point to the source; we don't replace it.
MLA compression reduced KV cache requirements by over 90% across DeepSeek's model generations.
60% confidenceConfirmed more than $1 trillion in combined Blackwell and Rubin purchase orders through 2027.
60% confidenceDeepSeek-V4-Pro needs roughly 10% of the memory that V3.2 required for million-token conversations.
60% confidenceThe Groq 3 LPX and Vera Rubin NVL72 combination delivers up to 35 times higher inference throughput per megawatt for trillion-parameter models.
60% confidenceStock Advisor's total average return is 902%, a market-crushing outperformance compared to 207% for the S&P 500.
60% confidenceThe AWS-Cerebras disaggregated inference solution delivers inference an order of magnitude faster than existing GPU-only solutions.
60% confidencePer-user generation speed improved 60% to 85% on DeepSeek-V4-Flash and 57% to 78% on V4-Pro over the prior baseline.
60% confidenceConsensus Data Center gross margin for fiscal 2027 is projected at 76.3%, slightly below fiscal 2024 and 2025 levels.
60% confidenceThroughput at a fixed service level improved 51%.
60% confidenceFull-year fiscal 2027 Data Center revenue consensus sits near $343 billion.
60% confidence
Data points we hold from this source
| Stock Advisor · market share | 902 percent_total_average_return |