$NBIS — $158.74
A waterfall that has finally reached oversold, support, and drying volume at the same time — the setup is better, the turn is not confirmed yet.
Support: $155–159, then $146–148
Holding $155–159 keeps this a washout that can base; losing it opens the July washout low at $146–148, and that break would end the starter thesis.
price closed Friday at $158.74, below all three moving averages — the 50-day at $217.84, the 100-day at $219.78, and the 200-day at $175.88. Those three values match chart's tags exactly, so there is no guesswork in them. The fall is now 47 percent from the $299.86 June high, 38 percent from the October high of $255.10, and 24 percent just since the October 12 breakdown. Two weeks ago the read at $202.74 was to wait, and the $197–200 support it rested on has since broken — the wait was right, and this is a different price now.
What makes this level different is exhaustion, not strength. RSI sits at 30.2, right on the oversold line. Friday traded only 13.5 million shares against a 20.4 million 20-day average, and far below the 47.7 million capitulation day on October 15 — the selling is running out of sellers, which is what volume contraction on a decline means. The accumulation line has fallen to 3.03 billion from 3.26 billion on October 12, so distribution is real, but it has bounced off its capitulation low rather than making new ones. Friday itself printed a small doji, indecision after a straight fall. None of that is a reversal signal. It is the anatomy of a washout: oversold, supported, quiet.
The fundamental case is what the price is now catching down to. Second-quarter revenue was $582.3 million, up 454 percent year over year, with AI cloud revenue up 514 percent and annualized run-rate revenue at $3.0 billion, up 598 percent. Adjusted EBITDA margin was 41 percent for the group and 50 percent for the cloud segment, operating cash flow was about $2.2 billion, and cash stood at $8 billion. Full-year guidance never changed through the selloff: $3.0–3.4 billion of revenue and $7–9 billion of year-end ARR. Behind it sits more than $40 billion of contracted revenue from investment-grade customers — Microsoft at about $17.4 billion with an option to $19.4 billion, and Meta at $12 billion firm with a path to $27 billion. On October 1 the company raised on-demand GPU prices 17–21 percent, its second increase in three months, including on H100s, four-year-old silicon — demand outrunning supply across every generation. And the funding model is the structural edge over the debt-funded neoclouds: about 70 percent of second-quarter deals came with customer prepayments covering 50–60 percent of the build cost, with more than $9 billion of prepayments expected this year. Customers are financing the expansion. At $158.74 the market cap is about $43 billion against that $7–9 billion year-end ARR guide — the multiple compressed by nearly half while the guide, the contracts, and the pricing power did not move.
The risks stay on the table. Capital spending of $20–25 billion against $8 billion of cash means funding is a live question, customer concentration in two names is real, and short interest near 19 percent cuts both ways — fuel for a squeeze if the November 9 third-quarter print beats, and a warning that plenty of smart money still doubts the depreciation math. The street is split the same way: Rosenblatt at Buy and $304, Piper cut to $185 in the selloff.
this is a starter at support, not a full position. Start small while $155–159 holds, add on a reclaim of the 200-day at $175.88 with volume, and save the final tranche for a close above the old September base at $205–208. Below $155 on a closing basis, the thesis pauses and $146–148 decides whether it was a washout or something worse. Technically and fundamentally, the price finally pays you to begin — it does not yet pay you to be all in.