$NBIS Your one calculation is wrong: $45B divided by $180M is 250x. And trailing P/E is the wrong tool for a company that grew revenue 454% last quarter and sells out every GPU it brings online. Nebius guides $3-3.4B of revenue this year at a ~40% group adjusted EBITDA margin, roughly $1.2-1.4B of adjusted EBITDA, and every dollar goes straight back into capacity customers are fighting over. CFO Dado Alonso: "Once again, we sold out our capacity as demand continued to exceed available supply." CRO Marc Boroditsky: "4 or more customers competing for every GPU we bring online." What you skipped: • Q2 revenue up 454% to $582.3M and ARR up 598% to $3.0B, with a 49.7% adjusted EBITDA margin in AI cloud. Four landmark deals averaged over $1B each • 2026 guidance reaffirmed at $7-9B of ARR by year-end, after the AI cloud pipeline grew 3.5x in Q1 excluding hyperscaler deals • Over $40B of customer commitments. AAA-rated Microsoft is $17.4B, up to $19.4B, with every tranche delivered, and Meta is up to ~$27B • Over $9B of customer prepayments expected this year. 70% of Q2 deals included prepayments covering 50-60% of the capex, and payback fell to 1 year 10 months from 2-3 years • GPU prices raised 17-21% on October 1. H100 went from $3.85 to $4.50 an hour and B300 from $7.85 to $9.50 • 5 GW of contracted power targeted by year-end, including a 1.2 GW Pennsylvania campus whose first 250-300MW arrives by the end of 2027 • Rated Platinum in SemiAnalysis's ClusterMAX 3.0 alongside CoreWeave Now 2027. Microsoft's $17.4B runs five years out of Vineland, a site designed for up to 300MW. On 300MW that's about $11.6M per MW a year. Meta's $12B of dedicated Vera Rubin capacity, starting in early 2027, also runs five years, so $2.4B a year, or about 200MW at that rate. Volozh says Nebius plans to build more than 1 GW of new capacity a year starting in 2027. Take only 800MW of it: • 200MW for Meta: $2.4B • 600MW at $20M per MW, the bottom of Q2's $20-25M deals: $12.0B • Plus $7B, the bottom of the 2026 guide That's $21.4B of ARR at the end of 2027, and at Q2's 49.7% AI cloud margin, over $10B of adjusted EBITDA a year. Push 20% of the enterprise capacity into 2028 and ARR is still $19.0B. Wolfe models $13.9B, and BNP Paribas, with a $399 target, sees about $22B. That gives zero credit for the $40-50M per MW short-term deals Nebius has started signing, "and sometimes above" per Volozh, Token Factory, where inference workloads more than tripled in Q2, or being named Palantir's sovereign AI infrastructure partner. Volozh: "We could sell our entire 2027 capacity on these terms today." The CFO said they're choosing not to, "reflecting our confidence with respect to future pricing dynamics." At $45B, Nebius trades at 2.1x that $21.4B and 2.4x the stress case. The average analyst target is ~$284, and Rosenblatt is at $304. "Worthless": at June 30 Nebius had $10.34B of shareholders' equity and $8.04B of cash, plus a stake in ClickHouse, a company valued at ~$15B in January. Microsoft and Meta's commitments alone come to up to ~$46B, about the entire market cap you call worthless. Lenders priced its first secured loan, $775M, at SOFR + 2.50%, and investors bought $3.45B of its converts at a 0.50% coupon. They read the filings you didn't. On 1999: name one dot-com whose customers prepaid half the cost of its buildout, or that signed $17.4B with Microsoft and delivered every tranche. 27 years should be long enough to tell a company with no customers from one with over $40B of signed commitments. On "short any rally": anyone who shorted Nebius on P/E at $76.77 last December had lost $223 a share, nearly 3x what they sold it at, by the time it hit $299.86 in June. That's the trade you're recommending today. I'm long $NBIS.

Despite recent market volatility and the AI selloff, $NBIS remains my highest-conviction investment and is once again the largest holding in our portfolios. It’s rare to find a generational opportunity at the intersection of seemingly insatiable demand, low-cost financing, and management execution that continues accelerating despite the company’s growing scale. The Street expects roughly $16B in exit ARR for 2027. Following Q2, our full-company model projected approximately $31B. I increasingly believe even that may prove too low. Management has discussed short-term capacity deals approaching $50M in annualized revenue per MW. If Nebius can capture more of this demand through short-term auctions rather than locking capacity into lower-priced multiyear agreements, revenue per MW could meaningfully exceed our assumptions. Combined with accelerating capacity deployment, $NBIS could potentially exit 2027 at more than 2x current Street expectations. I’ve been deploying significant cash from our flagship and active Growth portfolios during this selloff. Looking beyond 2027, our models suggest $NBIS could surpass $100B in annual revenue by 2030. I increasingly believe even that extraordinary forecast could be exceeded. Two developments are changing my expectations. First, Vera Rubin could materially increase the revenue potential of newly deployed capacity, particularly in short- and medium-term contracts. Alongside Nebius’s expanding software stack, including Token Factory and zero-day model support, the company is moving further beyond commodity GPU rentals. Second, capacity acquisition is accelerating. Much of our original forecast depended on greenfield developments across the US and Europe coming online in 2028 and 2029. Management’s hybrid approach, incorporating brownfield acquisitions, colocation, and faster near-term deployments, could pull substantial revenue forward. Higher revenue per MW combined with faster capacity deployment creates enormous upside to our existing scenarios. New management announcements continue strengthening the probability of our bull case. Of course, none of this comes cheaply. Our CapEx forecasts are also substantially above Street expectations. But unlike the financing concerns recently surrounding $CRWV , Nebius benefits from a strong cash position, attractive debt terms, and significant flexibility across debt, convertibles, equity, and potential subsidiary monetizations. Management has demonstrated an impressive ability to balance these financing sources without compromising growth. I suspect the upcoming Q3 earnings could be a major catalyst if management provides formal 2027 ARR and CapEx guidance. Both figures could catch Wall Street badly off guard. The market continues to underestimate not only the scale of infrastructure Nebius is building, but also the increasingly valuable software platform sitting on top of it. $NBIS remains my highest-conviction long-term holding, and recent volatility has only strengthened that conviction.

Respectfully, this thinking is very common among AI bears, but gets the unit economics of AI almost perfectly backwards! With this thinking you'll get the Math wrong by orders of magnitude. Let me break down why: 1. "input costs are skyrocketing" input cost per gpu is definitely the wrong unit. what matters is cost per token, and that's collapsing because each new generation makes far more tokens from the same power. for example: $NVDA gb300 makes up to ~50x the tokens per MW of hopper. cost per 1M tokens went from $4.20 to $0.12. that's ~35x cheaper, even though each gpu "costs more". 2. "token prices are hyper-deflating" the falling prices are for old capability. new models make smarter tokens, and smarter tokens sell for more openai's cheapest model is $0.50 per 1M output tokens, its top model is $50. 100x apart. Quite important to understand that not all tokens are equal. 3. simple example: hopper: cost $4.20 per 1M tokens. sell at $10 gb300: cost $0.12 per 1M tokens, ~50x the tokens. now let's just for the sake of argument assume we sell at half the price, $5 profit per 1M tokens: hopper: $10 - $4.20 = $5.80 gb300: $5 - $0.12 = $4.88 about the same per token. but gb300 makes ~50x as many from the same power total profit: ~40x higher, with token prices down 50% cheaper tokens also unlock way more usage. google went from 9.7T tokens a month to 3.2 quadrillion in two years what you call the bear case is the most bullish setup in the industry.

Blackwell Ultra (B300) GPU rental prices up 17 consecutive days to a new all-time high, according to Compute Desk's index of verified transactions from U.S. neoclouds.

a rumor that the VARA AI fund has blown up is making the rounds

US CPI shows the stupidity of rate hikes. The entire impact comes from refined products. Rate hikes will not create more barrels nor increase refining, but they damage families and businesses.