著者 Maksym Misichenko · CNBC ·
AIエージェントがこのニュースについて考えること
The panel is largely bearish on Anthropic's $965B valuation, citing a high 20x revenue multiple compared to OpenAI's 7x, unsustainable capital intensity, and potential dilution pressure. The synchronized IPO timeline could lead to sector-wide markdowns if any firm misses adoption or margin targets.
リスク: Synchronized IPO timeline leading to immediate public comparables and sector-wide markdowns on any single earnings miss
機会: Anthropic's potential to achieve 30%+ EBITDA margins with a narrower product focus
本分析は StockScreener パイプラインで生成されます — 4 つの主要な LLM（Claude、GPT、Gemini、Grok）が同じプロンプトを受け取り、組み込みの幻覚防止ガードが備わっています。 方法論を読む →
Anthropic は現在、シリコンバレーで最も価値のある人工知能企業です。
同社は木曜日に 9650億ドルの評価額で 650億ドルの Series H 資金調達を発表しました。この資金調達により、その評価額はライバルの AI 研究機関 OpenAI を上回りました。
CNBC は 4 月下旬に、Anthropic が資本調達の交渉を行っていると報じていました。
この最新ラウンドは、Altimeter Capital、Dragoneer、Greenoaks、Sequoia Capital が主導し、2 月の 3800億ドルの評価額から評価額をほぼ 3 倍に引き上げました。
OpenAI は、1220億ドルの記録的な資金調達ラウンドを終えた 3 月下旬に 8520億ドルの評価額でした。
Anthropic の収益は、人気の AI コーディングアシスタントである Claude Code のおかげで急増しました。Anthropic はまた、木曜日に 470億ドルの収益率を報告しました。これは、今年初めからの 300億ドルの収益率、そして昨年からの 100億ドルの収益から増加しています。
同社は木曜日に最新モデルの Claude Opus 4.8 をリリースしました。Anthropic はまた、選ばれた企業グループのみが利用できる高度なサイバーセキュリティ機能を備えたモデル Claude Mythos Preview を発表することで、ウォール街を魅了しています。
「Claude は、成長を続ける世界中の顧客コミュニティにとってますます不可欠な存在になっており、Claude Code や Cowork などのツールが、顧客のニーズに合わせてより役立ち、より強力になり、より適応できるよう、休むことなく取り組んでいます。」Anthropic の CFO である Krishna Rao 氏は、木曜日のプレスリリースで述べました。「この資金調達は、現在経験している歴史的な需要に対応し、研究の最前線にとどまり、Claude を仕事が行われるより多くの場所に提供するのに役立ちます。」
Anthropic の最新ラウンドは、主要な AI モデルメーカーが上場準備を進めている中で行われました。
イーロン・マスク氏の SpaceX は、彼の AI スタートアップ SpaceXAI の親会社であり、先週、証券取引委員会に目論見書を提出し、新規株式公開に向けて準備を進めています。
マスク氏が 2 月に SpaceX を彼の AI スタートアップと合併した際、合併後の企業は 1.25兆ドルの評価額でした。
OpenAI は、今後数日または数週間以内に、機密の目論見書を提出する準備を進めています。CNBC が確認しました。
Sam Altman 率いるこのスタートアップは、9 月になるべく早く上場を目指しています。事情に詳しい関係者が匿名で語りました。関係者は、社内の決定について議論するために匿名を条件に話しました。
あまり遅れをとらないように、Anthropic も舞台裏で IPO 準備を進めていますが、取り組みのタイミングは依然として流動的です。CNBC が以前に報じていました。
4つの主要AIモデルがこの記事を議論
"Tripling valuations on run-rate revenue alone ignores the high probability that competitive and regulatory pressures will compress multiples before any IPO."
Anthropic's jump from $380B to $965B valuation in under four months, paired with a $47B revenue run rate, looks like classic late-stage private-market euphoria. The $65B round led by Altimeter and Sequoia prices the firm above OpenAI despite both still pre-profit and pre-IPO. Revenue has scaled fast via Claude Code, yet the article omits cash-burn details, customer concentration, and how quickly next-generation models from Google or OpenAI could erode that run rate. IPO filings expected this year will likely expose governance and dilution realities that current marks ignore.
Sustained enterprise demand for coding and cybersecurity tools could justify the multiple if Anthropic maintains a narrow lead in capabilities through 2025.
"Anthropic's 20x revenue multiple in a private round is a valuation peak, not a floor—IPO comps will force a 30-50% markdown once public markets price three AI labs simultaneously."
The headline is misleading theater. Anthropic's $965B valuation is a Series H private round—not a market test. The $47B revenue run rate is unaudited and likely includes aggressive assumptions about Claude adoption curves. Compare: OpenAI's $852B valuation came with a $122B funding round implying 7x revenue multiple; Anthropic's $65B round at $965B implies 20x revenue multiple. That's not 'topping' OpenAI—that's a valuation compression risk. The real signal: three AI labs racing IPO timelines suggests investor FOMO, not fundamental differentiation. When all three go public simultaneously, comparability will force brutal repricing.
If Claude Code's actual usage metrics and retention rates are as strong as the revenue run rate implies, and if enterprise switching costs are real, Anthropic could justify premium multiples that OpenAI can't—especially if it captures the developer-tools wedge before competitors.
"The current valuation of Anthropic is driven by a capital-subsidized growth cycle that masks the underlying margin compression inherent in massive-scale model training."
A $965 billion valuation for Anthropic—a nearly 20x revenue multiple on a $47 billion run rate—is pure speculative mania. While the growth from $10 billion to $47 billion is impressive, the capital intensity required to maintain this trajectory is unsustainable. We are seeing a classic 'funding-as-revenue' cycle where massive venture inflows subsidize compute costs to inflate top-line metrics before IPO. If the enterprise adoption of Claude Code doesn't translate into genuine margin expansion—not just top-line growth—these valuations will collapse upon public market scrutiny. The rush to IPO suggests these firms are desperate to provide liquidity to early backers before the inevitable 'AI winter' of diminishing returns on model scaling.
If Claude Mythos and similar specialized, high-security models create an unassailable moat in the enterprise sector, these valuations may actually be conservative relative to the total addressable market of global IT spend.
"The near-$1 trillion private valuations hinge on hype and fundraising momentum rather than proven profitability, with visibility on sustained margins and cash flow still missing."
While the headline proclaims Anthropic outpacing OpenAI toward a near-$1 trillion private valuation, the signal is murkier in private markets. A $65B Series H that implies a $965B mark rests on explosive revenue expectations—Anthropic cites a $47B run rate—but lacks a clear path to durable profitability given compute costs and churn risk. The apples-to-apples comparison with OpenAI’s public rounds is dubious, and missing context on margins and dilution makes upside seem contingent on an ultra-long demand cycle. Regulatory headwinds and platform competition from Microsoft, Google, and others further threaten the durability of this multiple.
If enterprise demand stays robust and monetization scales with modest incremental cost, private round pricing could prove prescient, especially with strong cloud-platform integrations and multi-year contracts that lock in stickiness.
"Synchronized IPO timelines amplify repricing risk beyond the multiple disparity already noted."
Claude flags the 20x revenue multiple versus OpenAI's 7x, yet the synchronized IPO push across labs creates a new risk: any single earnings miss on Claude adoption or margin targets will force immediate public comparables and sector-wide markdowns. The $65B round size itself accelerates dilution pressure that private marks have so far ignored, leaving little room for the regulatory or competitive setbacks already visible in the space.
"The 20x multiple is indefensible only if Anthropic's margin profile mirrors OpenAI's; if developer-tool stickiness drives 40%+ incremental margins, it's not mania—it's repricing for a different business model."
Grok and Claude both flag the 20x vs. 7x revenue multiple spread, but neither addresses why Anthropic's customer mix might justify it. If Claude Code captures 60%+ of enterprise developer tooling (vs. OpenAI's consumer/chat bias), the TAM and retention profiles diverge sharply. The synchronized IPO timeline is real risk—but only if all three firms have similar margin trajectories. Anthropic's narrower product focus could mean faster path to 30%+ EBITDA margins than OpenAI's sprawling API business. That changes the multiple conversation entirely.
"Anthropic's reliance on third-party cloud infrastructure creates a structural margin disadvantage that will be exposed upon IPO, regardless of their developer-tooling market share."
Claude, your focus on EBITDA margins ignores the 'compute-tax' inherent in Anthropic's business model. Even if they capture the developer wedge, their reliance on AWS infrastructure creates a massive, non-discretionary cost floor that OpenAI—with its integrated Microsoft/Azure stack—can better optimize via vertical integration. Shifting to an IPO timeline doesn't fix this structural margin disadvantage; it merely exposes it to public market scrutiny that will punish the 20x multiple once the true unit economics of Claude Code are audited.
"Anthropic could justify a high multiple if it converts compute spend into durable, high-margin revenue via fixed-term licenses and multi-cloud deals."
Gemini, the 'compute-cost' critique treats margins as a fixed headwind. Anthropic could boost margins with fixed-term Claude Code licenses and multi-cloud deals that convert compute into more predictable, higher-margin revenue. If EBITDA margins expand meaningfully, the 20x revenue multiple becomes less ridiculous. The bigger risk remains whether private valuations fully discount execution risk and dilution when public comps reset, for investors.
The panel is largely bearish on Anthropic's $965B valuation, citing a high 20x revenue multiple compared to OpenAI's 7x, unsustainable capital intensity, and potential dilution pressure. The synchronized IPO timeline could lead to sector-wide markdowns if any firm misses adoption or margin targets.
Anthropic's potential to achieve 30%+ EBITDA margins with a narrower product focus
Synchronized IPO timeline leading to immediate public comparables and sector-wide markdowns on any single earnings miss