Wɔ Dylan Patel, a ɔhyehyɛɛ SemiAnalysis, AI adwumayɛ ho amanneɛbɔ krataa ne nhwehwɛmu adwumakuw a ɛwɔ nkɛntɛnso no fam no, Nvidia fata sɛnea wɔhwɛ no denneennen na wosua tabloid nsɛm ho amanneɛbɔfo a bere bi na wɔde wɔn ho ama Ɔhemmaa Diana no. Ɔ hwehwɛ chip adwumakuw kɛse no nneɛma foforo, nwoma ne kyɛ ne scoops nsɛm a ɛfa ne mu adwumayɛ ho. Ne nsɛm a ɔde kɔ amannɔne no yɛ nea ennyae araa ma ɛtɔ mmere bi a ɔhyɛ adwumakuw no abufuw, a ɔkae sɛ bere bi wɔde no too blacklist fii ne nsɛm a esisi no mu bere tiaa bi akyi bere a otintim nea Nvidia bɛka ho asɛm wɔ nhyiam bi a ɛreba no bi akyi.
Nanso, ɔsram biako a atwam ni no, Patel a wadi mfe 29 no nyaa bere tiaa bi a nsoromma a ɛrepue biara a ofi atetesɛm nsɛm ho amanneɛbɔ wiase no mu ntumi nhwɛ kwan da: baguam nkamfo a efi Nvidia CEO Jensen Huang hɔ. Wɔ GTC a ɛtwa toɔ no mu, adwumakuo no developers’ nhyiamu a wɔhwɛɛ no kɛseɛ no, Huang hwehwɛɛ Patel din na ɔkaa SemiAnalysis foforɔ assessment of chip performance, InferenceX, a ɛmaa Nvidia nkontabuo a ɛkorɔn.
“Ɔkaa nnipa baanu din pɛ wɔ nkyerɛkyerɛmu no nyinaa mu: me de ne OpenClaw guy no de,” Patel kae, na ɔde anigye susuw nkamfo a efi SemiAnalysis’ adwumayɛbea foforo a ɛyɛ ahwehwɛ a owia hann wom wɔ San Francisco kurow no mfinimfini no ho. (Sɛ yɛbɛka no yiye a, Huang kaa afoforo kakraa bi nso ho asɛm ampa wɔ ne passing sides ne aseda mu, a Sequoia Capital’s Alfred Lin ka ho.) “Ɔde yɛn ahyɛnsode no too soro, na ɔkaa slide no ho asɛm simma anum —ɛno yɛ soronko koraa.” Na nanso, Patel kae sɛ wanfɛre sɛ obetintim krataa a enni anidaso a ɛfa Nvidia’s Rubin chips ho —amanneɛbɔ a ebia ɛremma adwumakuw no ani nnye —nna pii akyi. Ɔsi so dua sɛ saa gyinaesi no kyerɛ sɛ adɛfɛdɛfɛsɛm rensiw no kwan sɛ ɔbɛbɔ ɔsebɔ a ɛkame ayɛ sɛ ɛyɛ dɔla ɔpepepem 5.
