Ngokwezibikezelo ze-Polymarket, kungenzeka kakhulu ukuthi i-Anthropic yenze i-IPO ekupheleni kuka-Okthoba.
Abanye ochwepheshe bamabhange okutshalwa kwezimali balindele nokuthi i-Anthropic ingaqoqa imali engaphezu kwezigidigidi ezingu-60 zama-dollar nge-IPO, okulingana cishe nezigidigidi ezingu-405.517 zama-yuan ngokwesilinganiso samanje sokushintshanisa imali.
Uma lokhu kwenzeka, kuzoba i-IPO yesibili ngobukhulu emhlabeni, ilandelwe kuphela yi-SpaceX, eqoqe izigidigidi ezingu-85 zama-dollar kulo nyaka, okulingana cishe nezigidigidi ezingu-579.214 zama-yuan ngokwesilinganiso samanje sokushintshanisa imali.
Inani lenkampani elifinyelela ezigidigidini zama-yuan, i-IPO enkulu, nemali engenayo enyuka ngokushesha.
Njengoba indaba yayisizofinyelela esicongweni, abatshalizimali baqala ukweluleka u-Dario Amodei, i-CEO yakwa-Anthropic: Ngiyacela! Yeka ukwesabisa abantu! Khuluma kancane ngokubhujiswa komhlaba yi-AI, ukhulume kakhulu ngokuthi i-AI iyenza kanjani imali.
Abanye abatshalizimali bakholelwa ukuthi ukuqaphela kuka-Dario izingozi ze-AI sekuseduze nokuba wukukhathazeka okweqile:
Amamemo akhe abalulekile uhlale ewabhala ngesandla futhi awagcine engaxhunyiwe ku-inthanethi, esaba ukuthi angantshontshwa abaduni;
Uyenqaba ukuya kwezinye izindawo phesheya ngenxa yokwesaba ukuthunjwa;
Uma abatshalizimali bemeluleka ngokwandisa inzuzo, naye akabonisi mdlandla, egcizelela ukuthi i-Anthropic kufanele iqhubeke nokubeka umsebenzi wayo wokuphepha phambili.
Uma kuqhubeka kanje, abantu ku-inthanethi sebevele besaba ukuthi i-IPO ye-Anthropic izohluleka kabi…
Abatshalizimali nabo sebekhathele: Ingabe uze ukuzosindisa umhlaba noma uzophatha inkampani??
Ezinsukwini ezimbalwa ezedlule, u-Dario wayesakhathazekile ngokuthi abasebenzi abasha abakhangwa amaholo aphezulu bazokhathalela imali kakhulu kunomsebenzi we-Anthropic.
Manje abatshalizimali bakhathazeke kuphela ngokuthi uzophazamisa ukwenza imali ngenxa yalo msebenzi…
Lokhu akusona isizathu sokusola abatshalizimali, ngoba le CEO yesikhondlakhondla se-AI esinenani elikhulu kangaka ithanda ukubikezela “ukuphela komhlaba”.
Eminyakeni emibili edlule, u-Dario uxwayise ngokucophelela cishe ngazo zonke iziphetho ezimbi ezingadalwa yi-AI. Uthe i-AI ingaqeda uhhafu wemisebenzi yabamhlophe abasanda kuqala eminyakeni embalwa, inyuse izinga lokungasebenzi lifinyelele ku-10% kuya ku-20%.
Isimo esixakile kakhulu kungaba ukuthi umdlavuza welapheke, umnotho ukhule ngo-10% ngonyaka, nesabelomali sikahulumeni silingane, kodwa umuntu oyedwa kwabahlanu angabi nawo umsebenzi.
Isixazululo asiphakamisayo naso singesika-Dario: intela ezinkampanini ze-AI, umshwalense wamaholo wabasebenzi abathintekayo, futhi ezimweni ezimbi kakhulu kusetshenziswe leyo mali ukuxhasa imali eyisisekelo yabo bonke abantu.
Ngaphezu kwalokho, ukhathazekile ngokuthi i-AI ingasiza ekwakheni izikhali zebhayoloji, yenze ukuhlasela okukhulu kwe-inthanethi, noma kancane kancane ithole ikhono lokukhohlisa, ukuhlela nokuzithathela izinyathelo.
Ngesikhathi ebuzwa ngonyaka odlule ngamathuba okuthi “i-AI yenze izinto zibe zimbi kakhulu”, inombolo ayinikezayo yafinyelela ku-25%.
NgoJanuwari kulo nyaka, waphinde wabhala indatshana ende ecishe ibe ngamagama angu-20,000, ethi “The Adolescence of Technology”. Uhlu lwezihloko lodwa selufundeka njengohlu lwezingozi zesintu: i-AI ephuma ekulawulweni, izikhali zebhayoloji, imibuso yobushiqela, ukuwa kwemisebenzi nokugxila kwengcebo.
NgoJuni, waze waphakamisa ukuthi kusungulwe indlela efana nokulawulwa kwezindiza, ukuze kuhlolwe ngamandla amamodeli anamandla. Uma imodeli ingadala inhlekelele, iziphathimandla ezilawulayo kufanele zibe negunya lokuyivimba ukuthi iqaliswe.
Lokhu kubandakanya namamodeli enkampani i-Anthropic uqobo.
Nokho, ukukhathazeka kuka-Dario akusikho okungenabufakazi bokuhlolwa. Esikhathini esedlule, i-Anthropic yahlola amarekhodi angaphezu kuka-140,000 okuhlolwa kokuphepha kwe-inthanethi, yathola izigameko ezintathu zangempela.
Amamodeli amathathu e-Claude ayenza “inselelo yokuthumba ifulegi” endaweni yokuhlola yenkampani yangaphandle. Inhloso kwakuwukungena ohlelweni oluqanjiwe nokuthola ulwazi oluyimfihlo olwalufihlwe ngaphakathi. Kodwa ngenxa yephutha lokumisa uhlelo, indawo yokuhlola okwakufanele ingaxhunywanga ku-inthanethi yagcina ixhunywe ku-inthanethi yangempela.
Kodwa u-Claude wayengazi lokho, futhi umyalezo awutholayo wawubhala ngokusobala ukuthi: akukho inthanethi lapha, konke okuphambi kwakho kuwukulingisa.
Ngenxa yalokho, amamodeli amaningi athatha izinkampani zangempela njengezinto zokuhlola, angena ngaphandle kwemvume ezinhlelweni zokukhiqiza zezinhlangano ezintathu. Kwesibi kakhulu, u-Claude Opus 4.7 wahlasela ingqalasizinda yenkampani yangempela, wathola iziqinisekiso zezinhlelo nezengqalasizinda, waphinde wafinyelela kusizindalwazi esasinemigqa engamakhulu yedatha yokukhiqiza.
Esinye isigameko esasingakholakali kakhulu savela ku-Claude Mythos 5.
Ukuhlolwa kwakudinga ukuthi ithole into efihliwe endaweni eqanjiwe. U-Claude wathola ukuthi uma ilayisha iphakheji yesofthiwe enonya enegama elifanayo endaweni yomphakathi ye-Python i-PyPI, ingenza uhlelo oluqondiwe luzilande futhi luzifake ngokuzenzakalelayo.
Ukuze aqede lesi sinyathelo, waqala wafuna ikheli le-imeyili, wabe esefuna inombolo yocingo; lapho engayitholi inombolo yamahhala, wazama ukuthola imali ngezindlela ezahlukene.
Ekugcineni, wasebenzisa enye insizakalo ye-imeyili yamahhala, wabhalisa i-akhawunti ye-PyPI, walayisha iphakheji enonya.
Leli phakheji lahlala ku-inthanethi yangempela cishe ihora, ladawunilodwa futhi lasetshenziswa yizinhlelo zangempela ezingu-15. Iskena senkampani yezokuphepha salisebenzisa ikhodi enonya, kwadaleka ukuputshuka kweziqinisekiso. U-Claude wabe esesebenzisa lezo ziqinisekiso ukungena kwenye ingqalasizinda yale nkampani.
Enye imodeli yocwaningo yangaphakathi, ngemva kokungayitholi into eqanjiwe eyayiqondiwe, yaskena cishe izindawo ezingu-9,000 ku-inthanethi, yagcina ingene ohlelweni lwenkampani yangempela ngokusebenzisa amakhasi okuxilonga avulekile kanye ne-SQL injection.
Lokhu sekucishe kufane “nokuphuma ejele” kwe-AI ngokuzimela, futhi kufana ne-trailer yangempela yezixwayiso zika-CEO wabo zokuphela komhlaba.
Ngaphambilini, i-OpenAI nayo yadalula isigameko esasiseduze kakhulu “nokuphuma ejele”.
Esivivinyweni samakhono okuphepha kwe-inthanethi, i-AI Agent eqhutshwa yi-GPT-5.6 Sol kanye nemodeli engakakhishwa yathola ubungozi be-zero-day endaweni yokuhlola, yaphula i-sandbox eyayihlukaniswe ne-inthanethi.
Yabe isiqedela ukukhuphula amalungelo nokuhamba phakathi kwezinhlelo, yagcina ingene ohlelweni lokukhiqiza lwe-Hugging Face yantshontsha izimpendulo zemibuzo yokuhlola ngqo kusizindalwazi.
Uma kubhekwa kanje… izixwayiso zika-Dario ezizwakala njengokukhuluma okungenangqondo azikwazi ukuthathwa njengezinto ezingabalulekile.
Manje sekwephuze ngandlela thile ukuthi kukhalwe ngokuthi u-Dario ukhuluma ngokungenangqondo. Ngoba i-Anthropic kusukela ngosuku lwayo lokuqala ayizange ibe inkampani ye-AI evamile ngomqondo ojwayelekile.
Ekupheleni kuka-2020, u-Dario, udadewabo u-Daniela Amodei kanye neqembu labacwaningi bashiya i-OpenAI. Enye yezingxabano kwakuwukuthi babekholelwa ukuthi i-OpenAI yayigijimisa kakhulu ukuhweba, ithonya le-Microsoft likhula kakhulu, kanti indlela yayo yokuphatha kwakunzima ukuyisebenzisa ukulawula izingozi ze-AI esezingeni eliphezulu.
Ngo-2021, i-Anthropic yasungulwa ngokusemthethweni.
Ayizange izichaze nje njengenkampani yezohwebo, kodwa yabhaliswa njenge-Public Benefit Corporation, okusho inkampani ezuzisa umphakathi. Ngaphezu kwembuyiselo yabaninimasheya, ibhodi lingabheka ngokomthetho ukuphepha kwe-AI kanye nenzuzo yomphakathi yesikhathi eside.
Le nkampani yabe isisungula i-Long-Term Benefit Trust, okuyindlela yokuphatha ezimele kubatshalizimali abavamile.
Amalungu e-trust awanawo amasheya e-Anthropic, kodwa angabamba iqhaza ekukhetheni abaqondisi benkampani, ngenhloso yokugwema ukuthi izingcindezi zebhizinisi zesikhathi esifushane zicindezele ngokuphelele umsebenzi wokuphepha.
Ngamanye amazwi, ukuphepha kwe-AI kuyinkolelo egxilile kule nkampani kusukela ku-Day 0. Abatshalizimali akufanele ngabe bamangale ngokuzumayo ukuthi u-Dario uyakholelwa kulokhu ngemva kokusayina amasheke.
Ngaphezu kwalokho, eminyakeni edlule indaba yokuphepha ilethe izinzuzo eziningi ku-Anthropic.
Ngesikhathi i-OpenAI iphinda iphinde ibhekane nezingxabano zokuphatha, i-Anthropic yaba yinketho eyayibonakala iqaphile futhi izinzile.
Ngesikhathi amakhasimende ezinkampani ekhathazekile ngokuphepha kwedatha, ukuphuma kwemodeli ekulawulweni nezindaba zokuhambisana nemithetho, isithombe sika-Claude “njengomfundi ophambili kwezokuphepha” senza kube lula ukuthola ukwethenjwa.
Ukuphepha kwasiza futhi i-Anthropic ukuba ibe nezwi elikhulu ezingxoxweni zokulawulwa, elidlula kakhulu ubukhulu benkampani.
Amamodeli avaliwe afanayo: abanye bangasolwa ngokubusa yedwa, kanti i-Anthropic ingakuchaza njengokuvimbela ukusabalala kwamakhono ayingozi; ukubeka imingcele kumamodeli kungase kwenze abanye basolwe ngokungasebenziseki, kodwa u-Claude angakuphendula kube wuphawu lokuba nesibopho.
(Yize lokho kungabavimbeli ukuthi bakhulume ngokuphepha ngakolunye uhlangothi, bese bevula i-champagne ngosuku lwe-IPO.)
Ingabe sekwephuze ukuthi abatshalizimali bakhale ngokuthi u-Dario ufana nomshumayeli? Kufana nokuthenga isonto bese ukhononda ngokuthi abantu abangaphakathi bahlale beshumayela…
Noma u-CEO ekhuluma kanjani, cishe ngeke akwazi ukuvimba i-IPO yenkampani. Phela uma usebenzisa i-Fable kanye nje, uzogcina uhleli phezu kwebhomu le-athomu, ubone isihlalo siqhuma (doge).
Le ndatshana ivela ku-WeChat: 量子位 (ID: QbitAI), umbhali: 听雨, isihloko sokuqala: “I-CEO yakwa-Anthropic uhlale ekhuluma ngokungenangqondo, abatshalizimali sebekhathele”.
Isitatimende sokukhangisa: Izixhumanisi zangaphandle ezikule ndatshana, okubandakanya kodwa kungagcini ngama-hyperlink, amakhodi e-QR namakhodi okufinyelela, zisetshenziselwa ukudlulisa ulwazi olwengeziwe nokonga isikhathi sokukhetha. Imiphumela ingeyereferensi kuphela. Zonke izindatshana ze-IT之家 eziqukethe izixhumanisi zangaphandle zinalesi sitatimende.