2025-11-27 · 个股论点

解析$NBIS旗下Avride与Toloka的高增长潜力，指出市场低估其子公司价值。

涉及标的： $AMZN $CRWV $GOOGL $META $MSFT $NBIS $SHOP $TSLA $UBER

中文翻译

$NBIS 即将超越 $CRWV，原因如下：在 $TSLA、$AMZN、$GOOGL 和现代汽车中，Nebius [ $NBIS ] 拥有美国仅有的 4 家 FSD-LVL 4 级自动驾驶出租车公司之一：Avride。人们还不知道的是： $NBIS 拥有 Toloka，其同比增长 140%。 Toloka 被 Anthropic、Hugging Face、$AMZN、$MSFT、ServiceNow 和 $SHOP 用于训练人工智能模型，在杰夫·贝佐斯风险投资部门的最新一轮融资中，Toloka 估值约为 7 亿至 12 亿美元。 Nebius 拥有多数股权（约 60%），但市场错误定价了其子公司与其核心业务一同以极高速率增长的事实。虽然不如 Avride 或 Clickhouse 具有主导地位，但鉴于其增长率，Toloka 代表了 Nebius 投资论点中的重要组成部分。我们最近看到 Scale AI（估值 290 亿美元）被 $META 收购（技术上约 49%），以此了解市场规模，并且由于利益冲突担忧，$GOOGL 和 OpenAI 的客户流失，这增加了 Toloka 的价值。据报道，Meta 收购后，Google 和 OpenAI 切断了与 Scale 的联系。仅 Google 在 2025 年为 Scale 预算了 2 亿美元。即使捕获其中 10-20% 的转移支出，也将实质性提升 Toloka 的收入。管理层指引 2025 年收入为 50-70 亿美元。牛市情景假设他们达到 70 亿美元以上（上限或超过）。鉴于他们在 2024 年增长了 140%，如果企业交易转化，这并非不合理。计算如下：70 亿美元 × 25 倍 = 175 亿美元。若为 80 亿美元且乘数为 25-30 倍（若势头异常强劲），则接近 20-25 亿美元。但鉴于其增长率，我们可能会看到：基准情景（12-15 倍乘数）：7-10 亿美元（当前）-> 9-13 亿美元（1 年后）。牛市情景 20-25 倍：12-17.5 亿美元（当前）-> 15-25 亿美元（1 年后）。 2024 年收入同比增长 140% 至约 2640 万美元，2025 年指引为 5-7 亿美元（同比增长 90-165%）。在一般牛市情景下，我们可能在 1 年内看到额外约 14 亿美元的子公司价值添加到分部加总估值中。多年后，如果公司和市场增长，该价值可能单独带来 30 亿、60 亿、90 亿美元。简而言之，市场仍在将 Nebius 的子公司估值为缓慢增长、商品化的业务，而非以三位数速率高价值复利的业务。真正的超额收益来自投资者理解市场错误定价了 $NBIS 这一篮子超高速增长公司。 $NBIS 拥有多个三位数增长的子公司，受益于并行价值创造：每项业务独立扩展，但其进展强化了母公司的估值。

英文原文

$NBIS is about to leapfrog $CRWV, here's why: Among $TSLA, $AMZN, $GOOGL, and Hyundai, Nebius [ $NBIS ] owns 1 of only 4 FSD-LVL 4 US Robotaxi companies: Avride. What people also don't know is this: $NBIS owns Toloka, growing 140% Y/Y. Used by Anthropic, Hugging Face, $AMZN, $MSFT, Servicenow, and $SHOP for training artificial intelligence models, Toklo is likely valued around $700M-$1.2B in its latest round by Jeff Bezos's venture arm. Nebius owns a majority stake (roughly ~60%), but the market is mispricing how their subsidiary companies grows at extreme rates alongside their core business. While not as dominant as Avride or Clickhouse, Toloka represents a meaningful piece the Nebius investment thesis given their growth rate. We’ve seen Scale AI ( $29B ) get acquired by $META recently (technically ~49%) to get a sense of market size, and we’ve seen customers from $GOOGL and OpenAI defect due to conflict of interest concerns, given added value to Toloka.   Google and OpenAI reportedly cut ties with Scale after Meta's acquisition. Google alone had $200M budgeted for Scale in 2025. Even capturing 10-20% of that displaced spend would materially boost Toloka's revenue. Management guided $50-70M for 2025. The bull case assumes they hit $70M+ (the upper bound or exceed it). Given they grew 140% in 2024, this isn't unreasonable if enterprise deals convert. The math: $70M × 25x = $1.75B. At $80M with 25-30x (if momentum is exceptionally strong), you get toward $2-2.5B. But given their growth rate we’ll likely see:  Base case (12-15x multiple): $700M-$1B (currently) -> $900M-$1.3B 1Y forward. Bull Case 20-25x: $1.2-1.75B (currently) -> $1.5-2.5B 1Y forward Revenue grew 140% year-over-year in 2024 to approximately $26.4 million, with 2025 guidance of $50-70 million (90-165% Y/Y growth). And with a general bull-case sceanrio, we might see an additional ~$1.4B in subsidiary value added to Sum of Parts in 1 year time. In multiple years time, that value might bring in $3B, $6B, $9B alone if the both the company and market grows. In short, the market is still valuing Nebius’s subsidiaries as if they were slow-moving, commoditized businesses instead of high-value compounding at triple-digit rates. The true alpha from investors understanding that the market is mispricing $NBIS basket of hyper-growth companies. $NBIS with multiple triple-digit-growth subsidiaries benefits from parallel value creation: each business scales independently, but their progress reinforces the parent company's valuation.

在 X 上查看原推 ↗