Web Interstitial Ad Example

史詩級交易背後的財務工程

08-10-2025
a20251008aa1

AMD與ChatGPT母企OpenAI建立策略夥伴關係。消息一出,AMD當晚股價飆升超過25%,且未來12個月內可能再漲40%。這筆龐大交易或將成為CEO如何利用公司資本支出需求,免費獲得頂尖硬體設備的經典案例!若計劃如期進行,Sam Altman可能憑空創造出1000億美元的價值。

今次交易包括AMD向OpenAI發行長期股票認購權證,以每股0.01美元價格「購買」最高達1.6億股AMD股票,約為現有16億股流通股的10%。當然,此權證的行使條件是OpenAI分階段從AMD購買高達6GW的數據中心GPU,且AMD股價在幾年內達到600美元。如果未來幾年完成約800億至1000億美元的6GW GPU採購目標,AMD的市場佔比將從目前的4.5%躍升至明年超過10%,並在幾年後達到15%。

此計劃可能使AMD未來五年的獲利預期翻倍,並在2030年前將每股盈餘提高至20美元。屆時,AMD股價可能攀升至600美元以上,讓OpenAI執行其所有長期認購權證,以持有AMD 10%股權,屆時AMD市值將達1兆美元。這部分持股將價值約1000億美元,抵銷OpenAI向AMD支付的800億至1000億美元。

雖然計劃仍有變數,難以斷言一定能完美實現,但成功機率相當可觀,且可讓OpenAI從大規模資本支出中獲取最多1000億美元利潤。執行長Altman深知AMD獲得此合約的重要性以及對AMD股價的長期正面影響,並利用這點安排這筆龐大交易,對OpenAI有利。OpenAI無論如何都需要投入數百億美元資本支出,卻成功利用這筆合約迫使AMD給予價值數年後可能高達1000億美元的巨額股票選擇權。

交易結構設計對OpenAI和AMD雙方都有利,因為可能推動AMD股價及市值進一步攀升。樂觀預期下,五年內總市值甚至可望翻三倍。一天內25%大漲或五年內市值翻四倍,皆足以支持AMD以10%股權換取如此大規模長期成長。

AMD從此交易中,營收、獲利與股價都有機會翻倍甚至三倍成長。

在6GW GPU訂單裡,首批1GW會在2026年下半年部署。這1GW訂單需約70萬片MI-355X(單片功耗1.4kW);按每片折扣價2萬美元(官方價2.5萬美元)計算,AMD 2026年GPU銷售額將達140億美元。假設淨利率50%,將產生70億美元淨利,對應每股盈餘可增加4.3美元。此前市場對AMD 2026年EPS平均預估為6.04美元,此交易或助其飆升至10美元。若如此,AMD股價可達300美元(市盈率30倍),比目前210美元高出43%。

整體6GW訂單在五至六年內總銷售額約為800億至1000億美元,將成為AMD晶片能力的重大背書,直接或間接推升其營收與獲利。若AMD成功交付6GW數據中心GPU,市佔率將從現有4%提升至12%以上,市值將自3340億美元躍升至2030年的1兆美元。在此情況下,OpenAI持有的10%股權將價值1000億美元,是相當值得的交易。步上揚。12個月,其股價容易達到210美元,相當於2026年每股盈利的35倍。在明年銷售額成長率維持在30%以上的多頭情況下,其本益比可能達到40倍,即240美元。

An epic deal shows how OpenAI CEO can create US$100 billion value from its capex plan

While AMD jumps 25%+ tonight and may gain another 40% in the next 12 months and appear to be the biggest winner in US stock market tonight, this mega deal may become the textbook case of how a CEO can leverage its firm’s capital expenditure need to get top-notch hardware equipment free of charge!  If this plays out as per the plan, Sam Altman may have created US$100 billion out of thin air.

How OpenAI could gain US$100 billion to pay for that 6GW GPU

The AMD-OpenAI deal includes AMD issuing long-term stock warrant to OpenAI to ‘buy’ at US$0.01 only of up to 160 million AMD stocks, or 10% of its current outstanding 1.6 billion shares.  Of course, it is subject to OpenAI buying as much as 6GW datacentre GPUs from AMD in stages and AMD stock price hitting US$600 within a few years.  If the first threshold of buying 6GW of GPUs, roughly US$80-$100 billion total sales, over the next few years is done, AMD could see its market share jumps from 4.5% now to 10%+ next year and may reach 15% a few years later.

That may double AMD’s earnings forecast in next 5 years and possibly boost its EPS to US$20 by 2030.  By then, AMD stock may climb above US$600 to allow OpenAI to vet all its long-term call option and own 10% of AMD with a total market cap of US$1 trillion.  That holding would then worth US$100 billion for OpenAI and offset the US$80-$100 billion payment to AMD.

It is too early to conclude that this plan would work perfectly as there are a few moving parts but there is a reasonable chance to success and allow OpenAI to make a maximum of US$100 billion profit from its big capex plan.  CEO Altman understands the importance for AMD to get this contract and the long-term positive implication on AMD stock price and make use of this to arrange this mega deal to benefit his firm.  OpenAI needs to do this billion US$ capital expenditure anyway but successfully make use of this mega contract to force AMD to give them a big stock option that may worth US$100 billion in a few years.

The deal is structured in a way that mutually benefits OpenAI and AMD as it would likely boost AMD stock price and total valuations (market capitalization) like the 25% leap tonight and maybe even bigger gain of another 43% in the next 12 months.  Or, in the bull case, triple its total value in 5 years.  For a one-day 25% jump or quadrupling in 5 years, it justifies AMD giving 10% of itself if such huge long-term huge growth takes place.

AMD could see its total sales, earnings, & stock price double or triple from this deal

Of the 6GW GPU orders, the first 1GW will be deployed in 2H2026.  On that 1GW contract, it would consume roughly 700,000 MI-355X (1.4kw each); at a discount price of US$20,000 instead of list price US$25,000, it would give AMD US$14 billion GPU sales in 2026.  Assuming a 50% net margin, it would generate US$7 billion net profit or additional EPS of US$4.3 per share.  Before this week, market’s average forecast for AMD 2026 EPS was US$6.04 and this deal may turbocharge it to US$10.  If that happens, AMD stock can reach US$300 (30x 2026 EPS), or 43% higher than US$210 now.

The full 6GW contract would translate to roughly US$80-US$100 billion total sales over 5-6 years and would be a big endorsement and proof of AMD chip’s capability that would boost its sales & earnings both directly and indirectly.  If AMD pulls this off and successfully delivers the 6GW datacentre GPU, it would boost its market share from 4% now to 12% or more and its total market cap could jump further form US$334 billion to maybe US$1 trillion by 2030.  In such case, the 10% share for OpenAI is a good deal although it would worth US$100 billion.

陳俊文(Norman)是一位獨立投資策略師,擁有超過30年專業投資經驗,曾在家族辦公室、銀行、精品基金公司、財富管理公司工作。在其職業生涯中,管理過多種資產,包括全球股票投資組合、多元資產投資組合、基金中的基金和可轉換債券基金。Norman對全球金融市場及多種投資資產具深入了解,包括全球股票、債券(高等級、高收益、混合、可轉換)、衍生性商品(選擇權、認股權證、掉期等)以及各類多頭和多頭產品對沖基金。他經常接受媒體採訪,並曾擔任多家大專院校專業金融課程的客座講師。
陳俊文