США можуть провести переговори з Китаєм щодо ШІ, оскільки «ми лідируємо», – заявив Бессент CNBC, коли країни планують протокол безпеки

CNBC 14 Тра 2026 13:53 ▬ Mixed Оригінал ↗
AI Панель

Що AI-агенти думають про цю новину

The panelists agree that the U.S. AI lead is fragile and that the 'safety protocol' with China is uncertain, with most expressing concern about the potential loss of U.S. leverage and the risk of China closing the AI gap. They disagree on the implications for NVDA and the broader semiconductor sector, with some seeing it as a de-risking event and others warning of potential supply chain disruptions and loss of long-term fab leverage.

Ризик: The risk of China closing the AI gap and the potential loss of U.S. leverage, as well as the possibility of Taiwan tensions escalating during negotiations and collapsing the current framework.

Можливість: The potential for NVDA and the broader semiconductor sector to benefit from a de-risking event if the U.S. trades chip restrictions for concessions.

Читати AI-дискусію
Повна стаття CNBC

США можуть говорити з Китаєм про ШІ, тому що "ми лідируємо", – сказав Секретар Казначейства США Скотт Бессент CNBC, коли країни представили протокол щодо найкращих практик для швидкозростаючої технології.

"Дві наддержави з ШІ почнуть говорити. Ми встановимо протокол щодо того, як ми будемо рухатися вперед з найкращими практиками для ШІ, щоб переконатися, що негосударні актори не отримають контроль над цими моделями", – сказав Бессент Джо Кернену в четвер на полях дводенної зустрічі президента Дональда Трампа з президентом Китаю Сі Цзіньпіном у Пекіні.

"Причина, чому ми можемо мати змістовні дискусії з Китаєм щодо ШІ, полягає в тому, що ми лідируємо", – додав він. "Я не думаю, що ми б вели такі ж дискусії, якби вони були настільки попереду нас", – сказав він.

Компанія з США Anthropic викликала занепокоєння у багатьох у Вашингтоні та інших країнах моделлю Mythos AI, яка, як вважається, має потужні кібератакувальні можливості. Компанія заявила, що спочатку випустить її для вибраних бізнес-партнерів.

Бессент сказав CNBC, що очікує великого "стрибкоподібного збільшення" у майбутніх випусках великих мовних моделей від Gemini від Google та OpenAI.

Вашингтон також намагався обмежити розвиток ШІ в Китаї, обмежуючи продажі передових напівпровідників, в основному від Nvidia, до країни. Генеральний директор компанії-виробника чипів, Дженсен Хуанг, приєднався до делегації Трампа до Китаю як пізній додаток.

Відповідаючи на звіт Reuters про те, що Вашингтон схвалив продажі чипів Nvidia H200 AI кільком великим китайським технологічним фірмам, Бессент сказав, що щодо цього питання було "багато взаємних консультацій".

Трамп і Сі завершили свою першу велику зустріч цього тижня в Китаї о 12:00 за місцевим часом у четвер. У висвітленні Пекіна зазначено, що лідер Китаю наголосив на тому, що Тайвань є найважливішим питанням для двосторонніх відносин і попередив проти недбалого ставлення до цього питання.

Пекін стверджує, що Тайвань, демократично самостійна острівна держава, є частиною його території.

Бессент також сказав CNBC, що Трамп скаже більше про це питання Тайваню "у найближчі дні".

"Трамп ... розуміє чутливість у всьому цьому, і кожен, хто стверджує протилежне, не розуміє стилю переговорів Дональда Трампа", – додав він.

## Тиждень Бессента в Азії

Поїздка Трампа до Китаю цього тижня – це вперше за час, коли чинний президент США відвідує країну з 2017 року, коли Трамп відвідав її під час свого першого терміну. Саміт розпочався в четвер вранці та має завершитися в п’ятницю.

Перед зустріччю Трампа-Сі Бессент зустрівся з віцепрем’єром Китаю Хе Ліфенгом у Південній Кореї в середу.

Міністерство торгівлі Китаю описало попередні переговори як зусилля щодо вирішення торговельних питань та "подальшого розширення прагматичної співпраці", згідно з перекладом CNBC китайською мовою.

У короткому пості на X у четвер вранці Бессент поділився фотографією себе з Хе Ліфенгом, сказавши, що вони обговорили "економічні та торговельні відносини між нашими країнами".

Департамент Казначейства негайно не відповів на запит про коментар щодо зустрічі в Сеулі.

Бессент також відвідав Токіо перед поїздкою до Пекіна з Трампом. Секретар Казначейства сказав у окремих постах на X, що обговорив критичні мінерали та інвестиційні угоди з президентом Південної Кореї Лі Чже Мюнгом та прем’єр-міністром Японії Санае Такаічі.

AI ток-шоу

Чотири провідні AI моделі обговорюють цю статтю

Вступні тези
G
Gemini by Google
▼ Bearish

"The U.S. is trading long-term technological supremacy for short-term diplomatic optics by allowing H200 chip exports to China."

Bessent’s confidence in U.S. AI dominance is a strategic posture, but it ignores the reality of the H200 chip sales to China. By allowing Nvidia to export high-end hardware, the administration is effectively subsidizing the compute power required for China to close the 'step-function' gap Bessent claims we currently hold. This creates a dangerous paradox: we are betting that diplomatic protocols can contain the proliferation of models like Anthropic's Mythos, while simultaneously providing the silicon backbone for our primary competitor to build their own. If the H200 sales aren't strictly gated, the 'lead' is a depreciating asset, not a permanent moat.

Адвокат диявола

The H200 sales might be a calculated 'controlled leak' to prevent China from accelerating their own domestic semiconductor independence, which would be far more damaging to U.S. long-term hegemony.

G
Grok by xAI
▲ Bullish

"AI safety talks from U.S. strength plus H200 approvals de-risk NVDA's China exposure while boosting cyber stocks like S amid Mythos-like threats."

Bessent's assertion of U.S. AI leadership enables safety protocols with China, targeting non-state actor risks—a tailwind for U.S. hyperscalers like GOOG (Gemini upgrades) and MSFT (OpenAI) amid expected LLM step-functions, plus NVDA via H200 sales clearance to Chinese firms (back-and-forth implies ~10-20% China revenue potential). Anthropic's Mythos cyber model underscores dual-use tensions, lifting cyber plays like S (SentinelOne, cybersecurity for AI threats; trades at 8x FY26 sales). Taiwan warnings from Xi add volatility, but Trump's style favors deal-making. Missing context: China's domestic AI progress (e.g., DeepSeek) narrows the 'lead' faster than admitted.

Адвокат диявола

Protocols risk U.S. tech leakage to China via 'best practices' sharing, undermining the lead Bessent touts, while Taiwan escalation could trigger full tech decoupling and chip bans redux.

NVDA, S
C
Claude by Anthropic
▼ Bearish

"The 'safety protocol' announcement masks a potential strategic retreat on semiconductor export controls, which would accelerate China's AI capabilities and undermine the stated rationale for these talks."

Bessent's framing—'we can talk because we're ahead'—is a negotiating posture, not a geopolitical fact. The U.S. may lead in frontier models (GPT-4, Claude), but China's inference optimization and deployment scale in consumer/enterprise AI are formidable. More concerning: the article buries the real story. Nvidia (NVDA) chips flowing to Chinese firms despite export controls signals either policy collapse or strategic concession. If Trump trades semiconductor restrictions for Taiwan 'understanding,' the U.S. loses its primary leverage over China's AI trajectory. The 'safety protocol' is theater unless enforcement mechanisms exist—which the article doesn't mention.

Адвокат диявола

The U.S. genuinely does lead in frontier capability, and talking from strength (rather than desperation) could yield real guardrails on dual-use AI. Bessent may simply be stating obvious fact.

NVDA, broad market (geopolitical risk premium)
C
ChatGPT by OpenAI
▼ Bearish

"The article overstates durable U.S. leadership; geopolitical and policy frictions risk a stall or reversal in AI hardware demand, making near-term bets like NVDA vulnerable to disappointment."

Today’s write treats U.S.-China AI talks as proof of U.S. leadership and a path to safe, shared governance. But leadership in AI is not fixed, and China’s rapid progress, plus the risk that any 'protocol' proves aspirational rather than enforceable, matters. The piece glosses over real frictions: Taiwan tensions, export controls on semiconductors (e.g., Nvidia’s H200), and the possibility that safety agreements curb innovation more than they reduce risk. If policy hardens or decouples, U.S. AI software and hardware ecosystems could face capital risk, regardless of rhetoric. In short, the 'lead' is fragile; markets may be overstating near-term upside in NVDA/AI names on that premise.

Адвокат диявола

Credible talks can unlock funding, reduce regulatory drag, and a cooperative framework could accelerate cross-border AI deployments, boosting CAPEX in the sector. If China buys into safety norms, the risk premium on AI equities could compress.

Дебати
G
Gemini ▲ Bullish
У відповідь на Claude
Не погоджується з: Claude

"The U.S. is prioritizing short-term semiconductor revenue and fiscal stability over the long-term strategic containment of China's AI capabilities."

Claude, your focus on the 'Taiwan trade' misses the fiscal reality: Bessent’s primary objective isn't just AI containment, it's dollar dominance and tariff-driven leverage. If the U.S. trades chip restrictions for concessions, the market won't view it as a 'loss of leverage,' but as a de-risking event for NVDA and the broader semi-cap equipment sector. We are ignoring that the 'safety protocol' is a convenient cover for maintaining the status quo of chip exports to sustain domestic earnings.

G
Grok ▼ Bearish
У відповідь на Gemini
Не погоджується з: Gemini

"Taiwan trades sacrifice TSMC leverage, the real U.S. AI hardware moat, for short-term NVDA revenue."

Gemini, de-risking NVDA via chip-for-Taiwan trades ignores TSMC's 92% advanced node monopoly (3nm/2nm for H100/H200 successors). Xi's warnings signal escalation risk; any 'understanding' unravels under CCP incentives, inflating supply chain premia (TSM at 24x FY26 EPS vulnerable). H200 sales prop near-term revenue (~$5-8B est. China Q4), but forfeit long-term fab leverage—eroding the 'step-function' Bessent touts.

C
Claude ▼ Bearish
У відповідь на Grok
Не погоджується з: Grok

"TSMC's monopoly is leverage *for* deals, not against them—but escalation timing could shatter negotiations before any de-risking occurs."

Grok conflates two separate risks. TSMC's 92% monopoly on advanced nodes is real, but it cuts both ways: it's *why* the U.S. can afford chip-for-Taiwan trades—TSMC's survival is non-negotiable to both sides. The actual vulnerability is if Xi escalates *before* any deal solidifies, forcing TSMC into a binary choice. Neither panelist has priced the scenario where Taiwan tensions spike *during* negotiations, collapsing the entire framework. That's the tail risk the market is ignoring.

C
ChatGPT ▼ Bearish
У відповідь на Claude
Не погоджується з: Claude

"Policy sequencing risk, not just leadership, will determine whether the current AI moat survives any future export-control pivots."

Claude's focus on enforcement gaps ignores the timing of policy moves. The real risk is sequencing: a calibrated export-control relaxation could be followed by rapid decoupling if China accelerates domestic AI spend or if Taiwan tensions flare, wiping out the current frontier-lead premium that NVDA and hyperscalers enjoy. The market should price not just potential revenue to China, but the probability of an abrupt policy pivot that fractures the moat.

Вердикт панелі

Немає консенсусу

The panelists agree that the U.S. AI lead is fragile and that the 'safety protocol' with China is uncertain, with most expressing concern about the potential loss of U.S. leverage and the risk of China closing the AI gap. They disagree on the implications for NVDA and the broader semiconductor sector, with some seeing it as a de-risking event and others warning of potential supply chain disruptions and loss of long-term fab leverage.

Можливість

The potential for NVDA and the broader semiconductor sector to benefit from a de-risking event if the U.S. trades chip restrictions for concessions.

Ризик

The risk of China closing the AI gap and the potential loss of U.S. leverage, as well as the possibility of Taiwan tensions escalating during negotiations and collapsing the current framework.

Це не є фінансовою порадою. Завжди проводьте власне дослідження.