AMD (AMD)

EarningsTranscriptsUpdated

As of Oct 24, 2026, Kalshi and Polymarket have settled 67 word markets on AMD’s earnings calls since 2025, on 49 different words. Next call: Nov 3, 2026 (16 words listed).

Next: Earnings call ·

16 words listed · Occasion page

Who counts: Kalshi: company speakers + operator

Kalshi

AMD: the words listed for Tue, Nov 3, how often each was said on past occasions, and the probability for this one
WordQ3’24Oct '24Q4’24Feb '25Q1’25May '25Q2’25Aug '25Q3’25year agoQ4’25Feb '26Q1’26May '26Q2’26last callTrendLast 8ScriptedAll-timeProbabilityMarket
Inference / Inferencing8 times: Inference / Inferencing, Q3’24 Oct '24, counts as said12 times: Inference / Inferencing, Q4’24 Feb '25, counts as said14 times: Inference / Inferencing, Q1’25 May '25, counts as said9 times: Inference / Inferencing, Q2’25 Aug '25, counts as said10 times: Inference / Inferencing, Q3’25 Nov '25, counts as said13 times: Inference / Inferencing, Q4’25 Feb '26, counts as said10 times: Inference / Inferencing, Q1’26 May '26, counts as said8 times: Inference / Inferencing, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%24/7831%95%n=1+tx from settlement history blended with transcripts78¢ Kalshi
Memory / HBM4 times: Memory / HBM, Q3’24 Oct '24, counts as said4 times: Memory / HBM, Q4’24 Feb '25, counts as said4 times: Memory / HBM, Q1’25 May '25, counts as said4 times: Memory / HBM, Q2’25 Aug '25, counts as said3 times: Memory / HBM, Q3’25 Nov '25, counts as said5 times: Memory / HBM, Q4’25 Feb '26, counts as said14 times: Memory / HBM, Q1’26 May '26, counts as said17 times: Memory / HBM, Q2’26 Aug '26, counts as saidrising8/8100%8/8100%36/7846%95%n=1+tx from settlement history blended with transcripts78¢ Kalshi
Helios0 times: Helios, Q3’24 Oct '240 times: Helios, Q4’24 Feb '250 times: Helios, Q1’25 May '258 times: Helios, Q2’25 Aug '25, counts as said14 times: Helios, Q3’25 Nov '25, counts as said10 times: Helios, Q4’25 Feb '26, counts as said8 times: Helios, Q1’26 May '26, counts as said21 times: Helios, Q2’26 Aug '26, counts as saidrising5/862%5/862%5/786%94%n=1+tx from settlement history blended with transcripts80¢ Kalshi
China2 times: China, Q3’24 Oct '24, counts as said0 times: China, Q4’24 Feb '256 times: China, Q1’25 May '25, counts as said7 times: China, Q2’25 Aug '25, counts as said2 times: China, Q3’25 Nov '25, counts as said10 times: China, Q4’25 Feb '26, counts as said6 times: China, Q1’26 May '26, counts as said1 time: China, Q2’26 Aug '26, counts as saidrising7/888%5/862%66/7885%90%n=2+tx from settlement history blended with transcripts68¢ Kalshi
OpenAI / ChatGPT0 times: OpenAI / ChatGPT, Q3’24 Oct '242 times: OpenAI / ChatGPT, Q4’24 Feb '25, counts as said1 time: OpenAI / ChatGPT, Q1’25 May '25, counts as said0 times: OpenAI / ChatGPT, Q2’25 Aug '2517 times: OpenAI / ChatGPT, Q3’25 Nov '25, counts as said5 times: OpenAI / ChatGPT, Q4’25 Feb '26, counts as said2 times: OpenAI / ChatGPT, Q1’26 May '26, counts as said4 times: OpenAI / ChatGPT, Q2’26 Aug '26, counts as saidrising6/875%6/875%9/7812%88%transcripts from the transcript model78¢ Kalshi
Microsoft / Azure7 times: Microsoft / Azure, Q3’24 Oct '24, counts as said4 times: Microsoft / Azure, Q4’24 Feb '25, counts as said1 time: Microsoft / Azure, Q1’25 May '25, counts as said1 time: Microsoft / Azure, Q2’25 Aug '25, counts as said3 times: Microsoft / Azure, Q3’25 Nov '25, counts as said1 time: Microsoft / Azure, Q4’25 Feb '26, counts as said0 times: Microsoft / Azure, Q1’26 May '266 times: Microsoft / Azure, Q2’26 Aug '26, counts as saidfalling7/888%7/888%29/7837%84%n=1+tx from settlement history blended with transcripts62¢ Kalshi
Agentic AI0 times: Agentic AI, Q3’24 Oct '241 time: Agentic AI, Q4’24 Feb '25, counts as said0 times: Agentic AI, Q1’25 May '251 time: Agentic AI, Q2’25 Aug '25, counts as said0 times: Agentic AI, Q3’25 Nov '250 times: Agentic AI, Q4’25 Feb '2610 times: Agentic AI, Q1’26 May '26, counts as said3 times: Agentic AI, Q2’26 Aug '26, counts as saidrising4/850%4/850%4/785%81%n=1+tx from settlement history blended with transcripts48¢ Kalshi
Meta5 times: Meta, Q3’24 Oct '24, counts as said2 times: Meta, Q4’24 Feb '25, counts as said1 time: Meta, Q1’25 May '25, counts as said0 times: Meta, Q2’25 Aug '252 times: Meta, Q3’25 Nov '25, counts as said0 times: Meta, Q4’25 Feb '263 times: Meta, Q1’26 May '26, counts as said4 times: Meta, Q2’26 Aug '26, counts as saidsteady6/875%6/875%10/7813%77%n=2+tx from settlement history blended with transcripts70¢ Kalshi
HPC / High-Performance Computing1 time: HPC / High-Performance Computing, Q3’24 Oct '24, counts as said4 times: HPC / High-Performance Computing, Q4’24 Feb '25, counts as said2 times: HPC / High-Performance Computing, Q1’25 May '25, counts as said1 time: HPC / High-Performance Computing, Q2’25 Aug '25, counts as said1 time: HPC / High-Performance Computing, Q3’25 Nov '25, counts as said3 times: HPC / High-Performance Computing, Q4’25 Feb '26, counts as said0 times: HPC / High-Performance Computing, Q1’26 May '261 time: HPC / High-Performance Computing, Q2’26 Aug '26, counts as saidfalling7/888%7/888%24/7831%76%n=1+tx from settlement history blended with transcripts46¢ Kalshi
Healthcare0 times: Healthcare, Q3’24 Oct '240 times: Healthcare, Q4’24 Feb '251 time: Healthcare, Q1’25 May '25, counts as said1 time: Healthcare, Q2’25 Aug '25, counts as said2 times: Healthcare, Q3’25 Nov '25, counts as said1 time: Healthcare, Q4’25 Feb '26, counts as said2 times: Healthcare, Q1’26 May '26, counts as said2 times: Healthcare, Q2’26 Aug '26, counts as saidrising6/875%6/875%15/7819%76%n=1+tx from settlement history blended with transcripts68¢ Kalshi
Oracle2 times: Oracle, Q3’24 Oct '24, counts as said0 times: Oracle, Q4’24 Feb '254 times: Oracle, Q1’25 May '25, counts as said2 times: Oracle, Q2’25 Aug '25, counts as said3 times: Oracle, Q3’25 Nov '25, counts as said0 times: Oracle, Q4’25 Feb '260 times: Oracle, Q1’26 May '261 time: Oracle, Q2’26 Aug '26, counts as saidfalling5/862%5/862%14/7818%69%n=1+tx from settlement history blended with transcripts46¢ Kalshi
Automotive / Automobile2 times: Automotive / Automobile, Q3’24 Oct '24, counts as said0 times: Automotive / Automobile, Q4’24 Feb '253 times: Automotive / Automobile, Q1’25 May '25, counts as said2 times: Automotive / Automobile, Q2’25 Aug '25, counts as said2 times: Automotive / Automobile, Q3’25 Nov '25, counts as said1 time: Automotive / Automobile, Q4’25 Feb '26, counts as said0 times: Automotive / Automobile, Q1’26 May '261 time: Automotive / Automobile, Q2’26 Aug '26, counts as saidfalling6/875%6/875%64/7882%68%n=1+tx from settlement history blended with transcripts46¢ Kalshi
Frontier1 time: Frontier, Q3’24 Oct '24, counts as said1 time: Frontier, Q4’24 Feb '25, counts as said2 times: Frontier, Q1’25 May '25, counts as said3 times: Frontier, Q2’25 Aug '25, counts as said0 times: Frontier, Q3’25 Nov '250 times: Frontier, Q4’25 Feb '260 times: Frontier, Q1’26 May '262 times: Frontier, Q2’26 Aug '26, counts as saidfalling5/862%5/862%9/7812%62%n=2+tx from settlement history blended with transcripts46¢ Kalshi
Export Restriction / Export Control0 times: Export Restriction / Export Control, Q3’24 Oct '240 times: Export Restriction / Export Control, Q4’24 Feb '254 times: Export Restriction / Export Control, Q1’25 May '25, counts as said7 times: Export Restriction / Export Control, Q2’25 Aug '25, counts as said0 times: Export Restriction / Export Control, Q3’25 Nov '250 times: Export Restriction / Export Control, Q4’25 Feb '260 times: Export Restriction / Export Control, Q1’26 May '261 time: Export Restriction / Export Control, Q2’26 Aug '26, counts as saidfalling3/838%3/838%6/788%49%n=1+tx from settlement history blended with transcripts46¢ Kalshi
TSMC0 times: TSMC, Q3’24 Oct '240 times: TSMC, Q4’24 Feb '252 times: TSMC, Q1’25 May '25, counts as said0 times: TSMC, Q2’25 Aug '250 times: TSMC, Q3’25 Nov '250 times: TSMC, Q4’25 Feb '260 times: TSMC, Q1’26 May '260 times: TSMC, Q2’26 Aug '26falling1/812%1/812%29/7837%15%n=2+tx from settlement history blended with transcripts24¢ Kalshi
Nvidia0 times: Nvidia, Q3’24 Oct '240 times: Nvidia, Q4’24 Feb '250 times: Nvidia, Q1’25 May '250 times: Nvidia, Q2’25 Aug '250 times: Nvidia, Q3’25 Nov '250 times: Nvidia, Q4’25 Feb '260 times: Nvidia, Q1’26 May '260 times: Nvidia, Q2’26 Aug '26steady0/80%0/80%7/789%13%n=2+tx from settlement history blended with transcripts46¢ Kalshi
  • times said on that call
  • counted as said (ring)
  • no call
  • probability next time, with n
  • Select a word or a cell for its quotes
What the columns mean
Last N
of the appearances shown, how many counted as said
Trend
how the count moved over these calls
Scripted
calls of the last 8 where it was said in the prepared remarks
All-time
every call on file
Probability
the calibrated probability for the next listing; n is the settled observations behind it when it was computed, '+tx' adds the transcripts, 'new' means too little history

Market prices are mid quotes from Kalshi and Polymarket, shown for context, may be delayed by several minutes and are not a signal. Quotes as of Oct 22, 2026 23:09 UTC.

Every word

1–36 of 36, most settled first
AMD: every word with settled markets, most settled first
WordQ3’24Oct '24Q4’24Feb '25Q1’25May '25Q2’25Aug '25Q3’25Nov '25Q4’25Feb '26Q1’26May '26Q2’26last callTrendLast 8ScriptedAll-timeProbability
HPC0 times: HPC, Q3’24 Oct '244 times: HPC, Q4’24 Feb '25, counts as said1 time: HPC, Q1’25 May '25, counts as said1 time: HPC, Q2’25 Aug '25, counts as said0 times: HPC, Q3’25 Nov '252 times: HPC, Q4’25 Feb '26, counts as said0 times: HPC, Q1’26 May '261 time: HPC, Q2’26 Aug '26, counts as saidfalling5/862%5/862%16/7821%64%n=2+tx from settlement history blended with transcripts
ROCm4 times: ROCm, Q3’24 Oct '24, counts as said6 times: ROCm, Q4’24 Feb '25, counts as said5 times: ROCm, Q1’25 May '25, counts as said7 times: ROCm, Q2’25 Aug '25, counts as said12 times: ROCm, Q3’25 Nov '25, counts as said3 times: ROCm, Q4’25 Feb '26, counts as said2 times: ROCm, Q1’26 May '26, counts as said9 times: ROCm, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%15/7819%94%n=2+tx from settlement history blended with transcripts
AI chipnever said0 times: AI chip, Q3’24 Oct '240 times: AI chip, Q4’24 Feb '250 times: AI chip, Q1’25 May '250 times: AI chip, Q2’25 Aug '250 times: AI chip, Q3’25 Nov '250 times: AI chip, Q4’25 Feb '260 times: AI chip, Q1’26 May '260 times: AI chip, Q2’26 Aug '26steady0/80%0/80%0/780%13%n=2+tx from settlement history blended with transcripts
AI / Artificial Intelligence20+ times48 times: AI / Artificial Intelligence 20+ times, Q3’24 Oct '24, counts as said52 times: AI / Artificial Intelligence 20+ times, Q4’24 Feb '25, counts as said43 times: AI / Artificial Intelligence 20+ times, Q1’25 May '25, counts as said46 times: AI / Artificial Intelligence 20+ times, Q2’25 Aug '25, counts as said57 times: AI / Artificial Intelligence 20+ times, Q3’25 Nov '25, counts as said56 times: AI / Artificial Intelligence 20+ times, Q4’25 Feb '26, counts as said61 times: AI / Artificial Intelligence 20+ times, Q1’26 May '26, counts as said69 times: AI / Artificial Intelligence 20+ times, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%14/7818%88%n=1+tx from settlement history blended with transcripts
Automotive2 times: Automotive, Q3’24 Oct '24, counts as said0 times: Automotive, Q4’24 Feb '253 times: Automotive, Q1’25 May '25, counts as said2 times: Automotive, Q2’25 Aug '25, counts as said2 times: Automotive, Q3’25 Nov '25, counts as said1 time: Automotive, Q4’25 Feb '26, counts as said0 times: Automotive, Q1’26 May '261 time: Automotive, Q2’26 Aug '26, counts as saidfalling6/875%6/875%64/7882%69%n=1+tx from settlement history blended with transcripts
Cloud28 times: Cloud, Q3’24 Oct '24, counts as said15 times: Cloud, Q4’24 Feb '25, counts as said12 times: Cloud, Q1’25 May '25, counts as said18 times: Cloud, Q2’25 Aug '25, counts as said12 times: Cloud, Q3’25 Nov '25, counts as said12 times: Cloud, Q4’25 Feb '26, counts as said10 times: Cloud, Q1’26 May '26, counts as said16 times: Cloud, Q2’26 Aug '26, counts as saidfalling8/8100%8/8100%41/7853%96%n=1+tx from settlement history blended with transcripts
CPU20+ times19 times: CPU 20+ times, Q3’24 Oct '24, below the threshold19 times: CPU 20+ times, Q4’24 Feb '25, below the threshold20 times: CPU 20+ times, Q1’25 May '25, counts as said20 times: CPU 20+ times, Q2’25 Aug '25, counts as said20 times: CPU 20+ times, Q3’25 Nov '25, counts as said39 times: CPU 20+ times, Q4’25 Feb '26, counts as said78 times: CPU 20+ times, Q1’26 May '26, counts as said35 times: CPU 20+ times, Q2’26 Aug '26, counts as saidrising6/875%0/80%10/7813%77%n=1+tx from settlement history blended with transcripts
Data center47 times: Data center, Q3’24 Oct '24, counts as said46 times: Data center, Q4’24 Feb '25, counts as said34 times: Data center, Q1’25 May '25, counts as said20 times: Data center, Q2’25 Aug '25, counts as said30 times: Data center, Q3’25 Nov '25, counts as said36 times: Data center, Q4’25 Feb '26, counts as said30 times: Data center, Q1’26 May '26, counts as said46 times: Data center, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%59/7876%97%n=1+tx from settlement history blended with transcripts
Data Center10+ times55 times: Data Center 10+ times, Q3’24 Oct '24, counts as said52 times: Data Center 10+ times, Q4’24 Feb '25, counts as said39 times: Data Center 10+ times, Q1’25 May '25, counts as said24 times: Data Center 10+ times, Q2’25 Aug '25, counts as said35 times: Data Center 10+ times, Q3’25 Nov '25, counts as said38 times: Data Center 10+ times, Q4’25 Feb '26, counts as said33 times: Data Center 10+ times, Q1’26 May '26, counts as said54 times: Data Center 10+ times, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%38/7849%90%n=1+tx from settlement history blended with transcripts
Desktop9 times: Desktop, Q3’24 Oct '24, counts as said6 times: Desktop, Q4’24 Feb '25, counts as said11 times: Desktop, Q1’25 May '25, counts as said8 times: Desktop, Q2’25 Aug '25, counts as said1 time: Desktop, Q3’25 Nov '25, counts as said4 times: Desktop, Q4’25 Feb '26, counts as said5 times: Desktop, Q1’26 May '26, counts as said1 time: Desktop, Q2’26 Aug '26, counts as saidfalling8/8100%7/888%31/7840%88%n=1+tx from settlement history blended with transcripts
Embedded14 times: Embedded, Q3’24 Oct '24, counts as said13 times: Embedded, Q4’24 Feb '25, counts as said14 times: Embedded, Q1’25 May '25, counts as said9 times: Embedded, Q2’25 Aug '25, counts as said10 times: Embedded, Q3’25 Nov '25, counts as said16 times: Embedded, Q4’25 Feb '26, counts as said14 times: Embedded, Q1’26 May '26, counts as said16 times: Embedded, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%48/7862%96%n=1+tx from settlement history blended with transcripts
EPYC20 times: EPYC, Q3’24 Oct '24, counts as said16 times: EPYC, Q4’24 Feb '25, counts as said22 times: EPYC, Q1’25 May '25, counts as said18 times: EPYC, Q2’25 Aug '25, counts as said16 times: EPYC, Q3’25 Nov '25, counts as said18 times: EPYC, Q4’25 Feb '26, counts as said15 times: EPYC, Q1’26 May '26, counts as said17 times: EPYC, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%20/7826%96%n=1+tx from settlement history blended with transcripts
Export restriction0 times: Export restriction, Q3’24 Oct '240 times: Export restriction, Q4’24 Feb '250 times: Export restriction, Q1’25 May '252 times: Export restriction, Q2’25 Aug '25, counts as said0 times: Export restriction, Q3’25 Nov '250 times: Export restriction, Q4’25 Feb '260 times: Export restriction, Q1’26 May '260 times: Export restriction, Q2’26 Aug '26steady1/812%1/812%2/783%26%n=1+tx from settlement history blended with transcripts
Game / Gaming5+ times13 times: Game / Gaming 5+ times, Q3’24 Oct '24, counts as said19 times: Game / Gaming 5+ times, Q4’24 Feb '25, counts as said24 times: Game / Gaming 5+ times, Q1’25 May '25, counts as said23 times: Game / Gaming 5+ times, Q2’25 Aug '25, counts as said14 times: Game / Gaming 5+ times, Q3’25 Nov '25, counts as said14 times: Game / Gaming 5+ times, Q4’25 Feb '26, counts as said18 times: Game / Gaming 5+ times, Q1’26 May '26, counts as said11 times: Game / Gaming 5+ times, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%21/7827%87%n=1+tx from settlement history blended with transcripts
Gaming12 times: Gaming, Q3’24 Oct '24, counts as said19 times: Gaming, Q4’24 Feb '25, counts as said23 times: Gaming, Q1’25 May '25, counts as said20 times: Gaming, Q2’25 Aug '25, counts as said13 times: Gaming, Q3’25 Nov '25, counts as said13 times: Gaming, Q4’25 Feb '26, counts as said18 times: Gaming, Q1’26 May '26, counts as said11 times: Gaming, Q2’26 Aug '26, counts as saidsteady8/8100%8/8100%25/7832%96%n=1+tx from settlement history blended with transcripts
GPU17 times: GPU, Q3’24 Oct '24, counts as said24 times: GPU, Q4’24 Feb '25, counts as said24 times: GPU, Q1’25 May '25, counts as said21 times: GPU, Q2’25 Aug '25, counts as said21 times: GPU, Q3’25 Nov '25, counts as said17 times: GPU, Q4’25 Feb '26, counts as said20 times: GPU, Q1’26 May '26, counts as said6 times: GPU, Q2’26 Aug '26, counts as saidfalling8/8100%8/8100%34/7844%96%n=1+tx from settlement history blended with transcripts
GPU10+ times28 times: GPU 10+ times, Q3’24 Oct '24, counts as said34 times: GPU 10+ times, Q4’24 Feb '25, counts as said32 times: GPU 10+ times, Q1’25 May '25, counts as said26 times: GPU 10+ times, Q2’25 Aug '25, counts as said28 times: GPU 10+ times, Q3’25 Nov '25, counts as said22 times: GPU 10+ times, Q4’25 Feb '26, counts as said26 times: GPU 10+ times, Q1’26 May '26, counts as said9 times: GPU 10+ times, Q2’26 Aug '26, below the thresholdfalling7/888%4/850%19/7824%74%n=1+tx from settlement history blended with transcripts
Graphic2 times: Graphic, Q3’24 Oct '24, counts as said2 times: Graphic, Q4’24 Feb '25, counts as said2 times: Graphic, Q1’25 May '25, counts as said0 times: Graphic, Q2’25 Aug '252 times: Graphic, Q3’25 Nov '25, counts as said0 times: Graphic, Q4’25 Feb '261 time: Graphic, Q1’26 May '26, counts as said2 times: Graphic, Q2’26 Aug '26, counts as saidfalling6/875%6/875%27/7835%71%n=1+tx from settlement history blended with transcripts
Inference4 times: Inference, Q3’24 Oct '24, counts as said8 times: Inference, Q4’24 Feb '25, counts as said3 times: Inference, Q1’25 May '25, counts as said5 times: Inference, Q2’25 Aug '25, counts as said10 times: Inference, Q3’25 Nov '25, counts as said13 times: Inference, Q4’25 Feb '26, counts as said5 times: Inference, Q1’26 May '26, counts as said3 times: Inference, Q2’26 Aug '26, counts as saidsteady8/8100%6/875%24/7831%94%n=1+tx from settlement history blended with transcripts
Inventory3 times: Inventory, Q3’24 Oct '24, counts as said6 times: Inventory, Q4’24 Feb '25, counts as said11 times: Inventory, Q1’25 May '25, counts as said9 times: Inventory, Q2’25 Aug '25, counts as said0 times: Inventory, Q3’25 Nov '255 times: Inventory, Q4’25 Feb '26, counts as said1 time: Inventory, Q1’26 May '26, counts as said2 times: Inventory, Q2’26 Aug '26, counts as saidfalling7/888%7/888%74/7895%80%n=1+tx from settlement history blended with transcripts
MI3008 times: MI300, Q3’24 Oct '24, counts as said8 times: MI300, Q4’24 Feb '25, counts as said5 times: MI300, Q1’25 May '25, counts as said5 times: MI300, Q2’25 Aug '25, counts as said3 times: MI300, Q3’25 Nov '25, counts as said0 times: MI300, Q4’25 Feb '260 times: MI300, Q1’26 May '260 times: MI300, Q2’26 Aug '26falling5/862%5/862%12/7815%54%n=1+tx from settlement history blended with transcripts
MI450 / MI455X0 times: MI450 / MI455X, Q3’24 Oct '240 times: MI450 / MI455X, Q4’24 Feb '250 times: MI450 / MI455X, Q1’25 May '250 times: MI450 / MI455X, Q2’25 Aug '2514 times: MI450 / MI455X, Q3’25 Nov '25, counts as said13 times: MI450 / MI455X, Q4’25 Feb '26, counts as said8 times: MI450 / MI455X, Q1’26 May '26, counts as said4 times: MI450 / MI455X, Q2’26 Aug '26, counts as saidrising4/850%4/850%4/785%91%n=1+tx from settlement history blended with transcripts
OpenAI0 times: OpenAI, Q3’24 Oct '242 times: OpenAI, Q4’24 Feb '25, counts as said1 time: OpenAI, Q1’25 May '25, counts as said0 times: OpenAI, Q2’25 Aug '2517 times: OpenAI, Q3’25 Nov '25, counts as said5 times: OpenAI, Q4’25 Feb '26, counts as said2 times: OpenAI, Q1’26 May '26, counts as said4 times: OpenAI, Q2’26 Aug '26, counts as saidrising6/875%6/875%9/7812%91%n=1+tx from settlement history blended with transcripts
PC / Personal Computer14 times: PC / Personal Computer, Q3’24 Oct '24, counts as said11 times: PC / Personal Computer, Q4’24 Feb '25, counts as said6 times: PC / Personal Computer, Q1’25 May '25, counts as said8 times: PC / Personal Computer, Q2’25 Aug '25, counts as said6 times: PC / Personal Computer, Q3’25 Nov '25, counts as said11 times: PC / Personal Computer, Q4’25 Feb '26, counts as said10 times: PC / Personal Computer, Q1’26 May '26, counts as said6 times: PC / Personal Computer, Q2’26 Aug '26, counts as saidfalling8/8100%8/8100%29/7837%94%n=1+tx from settlement history blended with transcripts
Supply Chain5 times: Supply Chain, Q3’24 Oct '24, counts as said0 times: Supply Chain, Q4’24 Feb '250 times: Supply Chain, Q1’25 May '256 times: Supply Chain, Q2’25 Aug '25, counts as said7 times: Supply Chain, Q3’25 Nov '25, counts as said4 times: Supply Chain, Q4’25 Feb '26, counts as said8 times: Supply Chain, Q1’26 May '26, counts as said6 times: Supply Chain, Q2’26 Aug '26, counts as saidrising6/875%2/825%46/7859%90%n=1+tx from settlement history blended with transcripts
Venice0 times: Venice, Q3’24 Oct '240 times: Venice, Q4’24 Feb '252 times: Venice, Q1’25 May '25, counts as said0 times: Venice, Q2’25 Aug '258 times: Venice, Q3’25 Nov '25, counts as said4 times: Venice, Q4’25 Feb '26, counts as said10 times: Venice, Q1’26 May '26, counts as said10 times: Venice, Q2’26 Aug '26, counts as saidrising5/862%5/862%5/786%91%n=1+tx from settlement history blended with transcripts
Xilinx0 times: Xilinx, Q3’24 Oct '240 times: Xilinx, Q4’24 Feb '251 time: Xilinx, Q1’25 May '25, counts as said0 times: Xilinx, Q2’25 Aug '250 times: Xilinx, Q3’25 Nov '251 time: Xilinx, Q4’25 Feb '26, counts as said0 times: Xilinx, Q1’26 May '260 times: Xilinx, Q2’26 Aug '26steady2/825%2/825%59/7876%36%n=1+tx from settlement history blended with transcripts
Buyback0 times: Buyback, Q3’24 Oct '240 times: Buyback, Q4’24 Feb '250 times: Buyback, Q1’25 May '250 times: Buyback, Q2’25 Aug '250 times: Buyback, Q3’25 Nov '250 times: Buyback, Q4’25 Feb '260 times: Buyback, Q1’26 May '260 times: Buyback, Q2’26 Aug '26steady0/80%0/80%12/7815%15%n=1+tx from settlement history blended with transcripts
Chiplet1 time: Chiplet, Q3’24 Oct '24, counts as said0 times: Chiplet, Q4’24 Feb '250 times: Chiplet, Q1’25 May '250 times: Chiplet, Q2’25 Aug '250 times: Chiplet, Q3’25 Nov '250 times: Chiplet, Q4’25 Feb '260 times: Chiplet, Q1’26 May '261 time: Chiplet, Q2’26 Aug '26, counts as saidsteady2/825%1/812%10/7813%34%n=1+tx from settlement history blended with transcripts
Custom silicon0 times: Custom silicon, Q3’24 Oct '241 time: Custom silicon, Q4’24 Feb '25, counts as said0 times: Custom silicon, Q1’25 May '250 times: Custom silicon, Q2’25 Aug '250 times: Custom silicon, Q3’25 Nov '250 times: Custom silicon, Q4’25 Feb '260 times: Custom silicon, Q1’26 May '260 times: Custom silicon, Q2’26 Aug '26steady1/812%0/80%3/784%19%n=1+tx from settlement history blended with transcripts
GPU shortagenever said0 times: GPU shortage, Q3’24 Oct '240 times: GPU shortage, Q4’24 Feb '250 times: GPU shortage, Q1’25 May '250 times: GPU shortage, Q2’25 Aug '250 times: GPU shortage, Q3’25 Nov '250 times: GPU shortage, Q4’25 Feb '260 times: GPU shortage, Q1’26 May '260 times: GPU shortage, Q2’26 Aug '26steady0/80%0/80%0/780%15%n=1+tx from settlement history blended with transcripts
MI300X8 times: MI300X, Q3’24 Oct '24, counts as said6 times: MI300X, Q4’24 Feb '25, counts as said1 time: MI300X, Q1’25 May '25, counts as said0 times: MI300X, Q2’25 Aug '253 times: MI300X, Q3’25 Nov '25, counts as said0 times: MI300X, Q4’25 Feb '260 times: MI300X, Q1’26 May '260 times: MI300X, Q2’26 Aug '26falling4/850%4/850%9/7812%42%n=1+tx from settlement history blended with transcripts
Milan1 time: Milan, Q3’24 Oct '24, counts as said2 times: Milan, Q4’24 Feb '25, counts as said0 times: Milan, Q1’25 May '250 times: Milan, Q2’25 Aug '250 times: Milan, Q3’25 Nov '250 times: Milan, Q4’25 Feb '261 time: Milan, Q1’26 May '26, counts as said0 times: Milan, Q2’26 Aug '26falling3/838%0/80%11/7814%30%n=1+tx from settlement history blended with transcripts
Pipeline0 times: Pipeline, Q3’24 Oct '240 times: Pipeline, Q4’24 Feb '250 times: Pipeline, Q1’25 May '250 times: Pipeline, Q2’25 Aug '250 times: Pipeline, Q3’25 Nov '250 times: Pipeline, Q4’25 Feb '261 time: Pipeline, Q1’26 May '26, counts as said0 times: Pipeline, Q2’26 Aug '26steady1/812%0/80%17/7822%22%n=1+tx from settlement history blended with transcripts
Redditnever said0 times: Reddit, Q3’24 Oct '240 times: Reddit, Q4’24 Feb '250 times: Reddit, Q1’25 May '250 times: Reddit, Q2’25 Aug '250 times: Reddit, Q3’25 Nov '250 times: Reddit, Q4’25 Feb '260 times: Reddit, Q1’26 May '260 times: Reddit, Q2’26 Aug '26steady0/80%0/80%0/780%15%n=1+tx from settlement history blended with transcripts
Triton2 times: Triton, Q3’24 Oct '24, counts as said1 time: Triton, Q4’24 Feb '25, counts as said0 times: Triton, Q1’25 May '250 times: Triton, Q2’25 Aug '251 time: Triton, Q3’25 Nov '25, counts as said0 times: Triton, Q4’25 Feb '260 times: Triton, Q1’26 May '260 times: Triton, Q2’26 Aug '26falling3/838%2/825%6/788%30%n=1+tx from settlement history blended with transcripts
What the columns mean
Last N
of the appearances shown, how many counted as said
Trend
how the count moved over these calls
Scripted
calls of the last 8 where it was said in the prepared remarks
All-time
every call on file
Probability
the calibrated probability for the next listing; n is the settled observations behind it when it was computed, '+tx' adds the transcripts, 'new' means too little history

Appearances and settled record

latest 12

Our transcript counts matched how 66 of 67 settled words resolved (98.5%). How we check.

Citeparrot.markets, “AMD (AMD): earnings call word history”, as of Oct 24, 2026, https://parrot.markets/earnings/AMD (CC BY 4.0)

Agents: this page as TSV at /api/v1/entity/amd · words: /api/v1/history?entity=amd · columnar JSON in #parrot-data · llms.txt · MCP