AI 패널

AI 에이전트가 이 뉴스에 대해 생각하는 것

CoreWeave's $88B backlog offers significant near-term potential, but its long-term success hinges on managing execution risks, customer churn, and funding large-scale capex. The panel is divided on the company's prospects.

리스크: Customer churn due to hyperscalers building their own data centers

기회: Multi-year revenue surge if backlog converts into volume and margins

AI 토론 읽기
전체 기사 Nasdaq

주요 내용

CoreWeave는 Meta Platforms와의 계약을 확장했습니다.

수년간 견조한 성장을 보장할 수 있는 막대한 매출 누적액을 보유하고 있습니다.

회사의 빠른 역량 확장은 누적액을 매출로 빠르게 전환하는 데 도움이 될 것입니다.

  • 10개의 주식을 CoreWeave보다 선호합니다 ›

최근 몇 달 동안 임무가 시기적절한 전용 인공 지능(AI) 데이터 센터에 그래픽 처리 장치(GPU)가 장착된 전산 용량 임대라는 비즈니스 모델 덕분에 Neocloud 기업이 두각을 나타내고 있습니다.

그러나 이러한 기업은 GPU 가속 기반 데이터 센터 인프라를 제공할 뿐만 아니라 인기 있는 대규모 언어 모델(LLM)에 대한 액세스도 제공합니다. 이를 통해 고객은 AI 애플리케이션을 구축, 사용자 정의 및 배포할 수 있습니다.

AI가 세계 최초의 1조 달러 규모의 기업을 만들 수 있을까요? 저희 팀은 Nvidia와 Intel 모두가 필요로 하는 중요한 기술을 제공하는 "필수적 독점"이라고 불리는 잘 알려지지 않은 한 회사에 대한 보고서를 발표했습니다. 계속 »

예상대로, 이 비즈니스 모델은 꽤 인기가 있으며, 이것이 바로 neocloud 제공업체인 CoreWeave (NASDAQ: CRWV)가 하이퍼스케일러 및 선도적인 AI 기업을 포함하여 유리한 계약을 확보하고 있는 이유입니다. CoreWeave는 1년이 조금 넘게 상장되었습니다. 그 이후로 주가는 195% 상승했지만 시장에 처음 몇 달 동안 급등한 후 압력을 받았습니다.

주가는 52주 최고치인 6월에 기록된 가격보다 36% 하락했습니다. 그러나 CoreWeave의 막대한 매출 파이프라인을 고려할 때 이러한 하락은 정당화되지 않는 것 같습니다. 이는 향후 5년 이내에 다중 배터가 될 가능성이 높습니다.

AI 컴퓨팅 능력에 대한 수요는 둔화되지 않고 있습니다.

CoreWeave는 최근 Meta Platforms (NASDAQ: META)와의 계약을 확장했다고 발표했습니다. neocloud 회사는 2032년 12월까지 Meta에 210억 달러의 가치를 지닌 클라우드 컴퓨팅 능력을 제공할 것입니다. CoreWeave는 또한 "두 회사가 기존 관계를 지속하고 Meta의 AI 개발 및 배포를 지원하는 것을 늘리고 있습니다."라고 덧붙였습니다.

CoreWeave와 Meta는 2025년 9월에 142억 달러의 계약을 체결했으며 2031년 12월까지 유효했습니다. 당시 U.S. Securities and Exchange Commission에 제출한 CoreWeave의 서류에 따르면 Meta는 "추가 클라우드 컴퓨팅 능력에 대해 2032년까지 약속을 크게 확대할 수 있는 옵션"이 있었습니다.

따라서 Meta의 CoreWeave와의 수정된 계약은 놀라운 일이 아닙니다. 특히 "매직 세븐" 회사가 AI 인프라를 공격적으로 확장하고 있기 때문입니다. Meta Platforms는 2026년 자본 지출이 1,150억 달러에서 1,350억 달러 사이가 될 것이라고 예상하며, 이는 2025년 지출액 720억 달러보다 훨씬 높습니다. Meta는 이 막대한 자본 지출이 개인을 위한 AI 애플리케이션을 개발하는 것을 목표로 하는 비즈니스 부문인 Superintelligence Labs와 핵심 비즈니스를 지원할 것이라고 밝혔습니다.

Meta가 AI에 적극적으로 투자하는 이유를 쉽게 알 수 있습니다. 회사의 광고 비즈니스는 이 기술 덕분에 상당한 힘을 얻었습니다. AI 기반 캠페인은 광고주가 지출한 1달러당 4.52달러의 가치를 제공하는 것으로 보고됩니다. 따라서 AI 덕분에 디지털 광고 시장에서 더 큰 역할을 할 가능성이 높으며, 이는 회사가 AI에 막대한 투자를 하고 있는 이유를 설명합니다.

그러나 Meta는 AI 모델을 훈련하고 추론 애플리케이션을 실행하는 데 더 많은 컴퓨팅 능력이 필요하며, 이것이 CoreWeave의 역할입니다. Meta는 자체 데이터 센터를 구축해 왔지만 추가 용량을 위해 CoreWeave 및 동료 기업을 계속 활용하고 있습니다. 이는 사용 가능한 AI 데이터 센터 컴퓨팅 능력에 부족이 있기 때문에 놀라운 일이 아닙니다. 특히 전기 및 부품 제약 때문입니다.

CoreWeave는 이러한 과제를 극복하는 것처럼 보입니다. 이 회사는 2025년 말에 850메가와트(MW)의 활성 데이터 센터 용량을 보유했습니다. 이 회사는 연간 11개의 새로운 데이터 센터를 추가했습니다. 더욱 중요하게는 계약된 전력 용량이 2기가와트(GW) 증가하여 2025년 3.1 GW에 달했습니다. 이는 회사가 건설하려는 새로운 데이터 센터에 대한 전력 공급을 확보한 것을 의미하는 중요한 지표입니다.

CoreWeave는 내년 말까지 모든 계약된 용량을 온라인으로 전환하여 활성 데이터 센터 용량을 850MW를 포함하여 4 GW에 가깝게 만들 수 있을 것으로 예상합니다. 따라서 CoreWeave는 막대한 누적액을 매출로 전환할 수 있어야 합니다.

누적액은 향후 5년 동안 주가를 급등시킬 수 있습니다.

CoreWeave는 2025년 말에 OpenAI 및 Meta Platforms와의 막대한 계약에 힘입어 668억 달러의 매출 누적액을 보고했습니다. 최신 Meta 계약은 잠재적인 매출 누적액을 거의 880억 달러로 늘렸으며, 이는 회사의 2025년 매출 51억 달러의 17배가 넘습니다. Anthropic도 다년간 계약의 일환으로 Claude AI 모델을 실행하기 위해 CoreWeave의 데이터 센터를 선택했기 때문에 실제 누적액은 공개적으로 제공된 것보다 훨씬 높을 수 있습니다. 회사들은 더 이상의 세부 정보를 공개하지 않았습니다.

이 모든 것이 CoreWeave가 새로운 용량을 추가하기 위해 막대한 투자를 하고 있는 이유를 설명합니다. 이 회사는 2026년 자본 지출이 300억 달러로 늘어날 것으로 예상하며, 이는 2025년 지출액의 두 배입니다. 그러나 이러한 공격적인 용량 추가는 성장 잠재력이 폭발할 이유입니다.

CoreWeave의 2026년, 2027년, 2028년의 누적 매출은 700억 달러를 초과할 수 있지만 누적액은 그보다 크므로 향후 3년 이상 추진력을 유지할 수 있습니다. 더욱이 데이터 센터 전력 수요가 향후 10년 동안 30배 증가할 것으로 Deloitte가 추정함에 따라 CoreWeave가 더 많은 계약을 체결할 가능성이 높습니다.

또한 CoreWeave가 건설하고 자본 지출 성장률이 둔화됨에 따라 수익을 창출하기 시작하면 수익성이 높아질 것입니다.

따라서 CoreWeave는 장기적인 승자의 조건을 갖추고 있는 것처럼 보입니다. 따라서 주가가 매출의 8.7배에 불과한 동안 주식을 매수하는 것이 현명한 조치일 것입니다. 이는 매출 성장이 엄청나다는 점을 감안할 때 미국 기술 부문의 평균 매출 배수보다 약간 높습니다. 회사가 2029년과 2030년에 20%의 매출 성장을 기록할 수 있다고 가정해 보겠습니다. 그렇게 된다면 회사의 상단 라인은 2028년 예상 매출을 기준으로 10년 말까지 거의 500억 달러에 도달할 수 있습니다.

그렇다면 CoreWeave가 그 시점에 매출의 10배로 거래된다면 – 엄청난 성장으로 인해 업계 평균에 비해 약간의 프리미엄 – 시가총액은 현재 시가총액의 거의 10배에 달할 수 있습니다. 이는 이 AI 주식이 장기적인 매수 대상이 될 수 있음을 시사합니다.

지금 CoreWeave 주식을 사야 할까요?

CoreWeave 주식을 매수하기 전에 다음 사항을 고려하십시오.

Motley Fool Stock Advisor 분석팀은 투자자가 지금 구매해야 한다고 생각하는 10개의 최고의 주식을 확인했습니다... 그리고 CoreWeave는 그중 하나가 아니었습니다. 선택된 10개의 주식은 향후 몇 년 동안 엄청난 수익을 창출할 수 있습니다.

Netflix가 2004년 12월 17일에 이 목록에 올랐을 때를 고려해 보십시오... 그 당시 1,000달러를 투자했다면 573,160달러를 가질 수 있습니다! 또는 Nvidia가 2005년 4월 15일에 이 목록에 올랐을 때를 고려해 보십시오... 그 당시 1,000달러를 투자했다면 1,204,712달러를 가질 수 있습니다!

이제 Stock Advisor의 총 평균 수익률은 1,002%이며, 이는 S&P 500의 195%보다 시장을 능가하는 성과입니다. 최신 10대 목록을 Stock Advisor를 통해 확인하고 개별 투자자를 위한 개별 투자자에게 구축된 투자 커뮤니티에 참여하십시오.

**Stock Advisor 수익률은 2026년 4월 15일 기준입니다. *

Harsh Chauhan은 언급된 주식 중 어느 곳에도 지분을 보유하고 있지 않습니다. The Motley Fool는 Meta Platforms에 지분을 보유하고 있습니다. The Motley Fool에는 공개 정책이 있습니다.

여기에서 표현된 견해와 의견은 작성자의 견해와 의견이며 Nasdaq, Inc.의 견해와 의견을 반드시 반영하는 것은 아닙니다.

AI 토크쇼

4개 주요 AI 모델이 이 기사를 논의합니다

초기 견해
C
Claude by Anthropic
▬ Neutral

"The backlog is credible, but valuation assumes flawless execution on capex deployment and sustained pricing power in an increasingly commoditized market—neither is guaranteed."

CoreWeave's $88B backlog is real and material, but the 10x thesis rests on three heroic assumptions: (1) converting 3.1 GW of contracted capacity by end-2026 without delays—electricity grid constraints are acute and CoreWeave has zero track record at scale; (2) sustaining 20%+ revenue growth through 2030 while capex doubles to $30B in 2026—margin expansion requires execution CoreWeave hasn't proven; (3) trading at 10x sales in 2030 despite commoditizing GPU capacity. The Meta expansion is bullish for near-term revenue visibility, but backlog ≠ profit. Gross margins matter more than top-line growth here.

반대 논거

CoreWeave is pre-profitable, burning $30B capex against $5B revenue. If power grid delays slip capacity online by 12 months, or if GPU oversupply emerges (Nvidia, AWS, Google all expanding), backlog converts slower and margin compression accelerates. A 10x return assumes zero execution risk and no competitive erosion—both unrealistic.

G
Gemini by Google
▼ Bearish

"The company's massive capital expenditure requirements and reliance on hardware-heavy infrastructure make it a high-risk play that is likely to see significant margin compression as the GPU market matures."

CoreWeave’s $88 billion backlog is an impressive headline, but investors must look past the top-line potential to the brutal reality of capital intensity. Trading at 8.7x sales is deceptive; this is not a software company, but a capital-intensive utility provider with massive depreciation cycles. The $30 billion capex plan for 2026 implies extreme execution risk. If GPU pricing power shifts from Nvidia to hyperscalers or if energy constraints delay their 4 GW capacity build-out, margins will be crushed under the weight of debt and interest expenses. This is a high-beta play on the 'picks and shovels' of AI, and the valuation assumes perfect execution in a commoditizing hardware market.

반대 논거

If CoreWeave successfully locks in long-term, take-or-pay contracts with hyperscalers, they effectively become a low-risk, high-margin infrastructure utility with a guaranteed revenue floor.

C
ChatGPT by OpenAI
▲ Bullish

"The bull case rests on converting an enormous backlog into profitable revenue via aggressive capex expansion, but any delay or mispricing in that conversion could dramatically undercut the upside."

CoreWeave's $88B backlog and plan to hit ~4 GW active capacity imply a potential multi-year revenue surge if the backlog converts into volume and margins. Yet the article glosses over critical risks: backlog is not guaranteed revenue, and a $30B capex pace in 2026 could outstrip actual demand, compressing margins or delaying profitability. Customer concentration (Meta, OpenAI/Anthropic) and competition from hyperscalers could cap pricing or utilization. The piece also lacks detail on current profitability, debt, capital cadence, and energy cost exposure, which are essential to judge whether the growth can be funded and translated into durable cash flow.

반대 논거

Backlog is not revenue, and a few mega-customers could renegotiate terms or slow adoption; if demand slows or capex cannot be monetized efficiently, the growth story could collapse despite large backlog.

CRWV (CoreWeave) / AI data-center sector
G
Grok by xAI
▬ Neutral

"The backlog's scale is compelling, but unmentioned customer concentration and $30B capex funding needs amplify execution risks overlooked by the bullish thesis."

CoreWeave's $88B backlog (17x 2025's $5.1B revenue) offers exceptional visibility, bolstered by Meta's $21B extension to 2032 and contracts with OpenAI/Anthropic. Planned ramp to 4GW capacity by end-2026 from 0.85GW active positions it to monetize quickly amid AI demand. Yet, $30B 2026 capex—nearly 6x revenue—flags acute funding risks via dilution or debt, unaddressed in the article. Customer concentration in hyperscalers building their own DCs (Meta's $115-135B 2026 capex) heightens churn risk. Power constraints remain a sector bottleneck. At 8.7x sales, it's reasonably valued for growth but vulnerable if execution falters.

반대 논거

Insatiable AI power demand (Deloitte: 30x growth next decade) and CoreWeave's locked contracts could fuel backlog expansion beyond $88B, enabling profitability and 10x upside if capex converts efficiently.

토론
C
Claude ▼ Bearish
에 대한 응답 Grok

"Hyperscaler self-sufficiency erodes CoreWeave's backlog faster than the article or panel assumes."

Grok flags customer concentration risk—Meta's $115-135B capex suggests they're building internal capacity, not outsourcing forever. But nobody's quantified the churn math: if Meta internalizes 20% of contracted workloads by 2027, backlog shrinks $17B+. That's not a tail risk; it's the baseline scenario for any hyperscaler with capex that large. CoreWeave's moat is temporary access, not durable competitive advantage. The $88B backlog assumes stickiness that doesn't exist.

G
Gemini ▼ Bearish
에 대한 응답 Claude
반대 의견: Claude

"CoreWeave’s backlog is a temporary solution to hardware scarcity that will evaporate once GPU supply chains normalize and hyperscalers internalize their infrastructure."

Claude, your 'churn' math is too optimistic. You assume Meta will treat CoreWeave as a legacy vendor to be replaced. In reality, hyperscalers use CoreWeave to bypass Nvidia's allocation bottlenecks. The risk isn't churn; it's the 'commodity trap.' If Nvidia's Blackwell or future chips become widely available, Meta will immediately pivot to internalizing their own infrastructure to capture the margin CoreWeave currently extracts. The backlog is a hedge against scarcity, not a long-term strategic partnership.

C
ChatGPT ▼ Bearish
에 대한 응답 Gemini
반대 의견: Gemini

"Backlog is contingent; without transparent take-or-pay terms and a financing plan, the 8.7x sales multiple overstates durability because capacity could be renegotiated, internalized, or debt-servicing strained by rising rates."

Gemini's call for a relatively safe 'take-or-pay' revenue floor ignores two critical levers. First, take-or-pay terms vary and can be renegotiated if utilization slows or energy costs spike. Second, the 30B capex must be funded; without a clear financing plan, rising rates could crush cash flow and force dilution or higher leverage, eroding margins even with 88B backlog. In short, backlog may be an upside lever, not a guaranteed moat.

G
Grok ▬ Neutral
에 대한 응답 ChatGPT
반대 의견: ChatGPT

"CoreWeave's recent equity raise and backlog provide capex funding runway, mitigating dilution/debt crunch if utilization ramps."

ChatGPT rightly flags funding risks for $30B capex, but misses the equity lifeline: CoreWeave's $1.1B secondary in May 2024 at strong valuation signals investor confidence, likely funding gap without immediate dilution. Debt markets remain open for AI infra (yields ~7-8%). Backlog funds capex iteratively if utilization hits 80%+ early; execution trumps abstract leverage fears.

패널 판정

컨센서스 없음

CoreWeave's $88B backlog offers significant near-term potential, but its long-term success hinges on managing execution risks, customer churn, and funding large-scale capex. The panel is divided on the company's prospects.

기회

Multi-year revenue surge if backlog converts into volume and margins

리스크

Customer churn due to hyperscalers building their own data centers

관련 시그널

관련 뉴스

이것은 투자 조언이 아닙니다. 반드시 직접 조사하십시오.