A Chinese AI developer pulling off a $5 billion Hong Kong raise shows global investors are still writing big checks for AI even as rate fear punishes US tech — meaning the sector's funding hasn't dried up, it's rotating. Oracle's earnings beat confirms th

Verdict: right theme, wrong week — wait for the sector gate to confirm

Verdict: wait for the trigger, not the story. The strongest point for this idea is that Alibaba's core economics remain mid-pack-to-good even in a down year — fiscal 2026 revenue grew 8.1% to $148.4B, gross margin held near 39.8%, and ROE of 9.8% ranks in the 70th percentile of 682 IT peers — so if Chinese AI sentiment heals, the discount is real. The strongest point against is that the very fundamentals are deteriorating while you wait: operating margin fell 9.24 points to 4.9%, operating cash flow halved to $11.0B, the dividend was cut to $1.05 from $2.00 (ex-date June 11, 2026), and six institutional filers showed net open-market selling of roughly $70.8M in the June 30, 2026 reporting window — a lagging filing, but directionally unhelpful. Mechanically, the trade is not close to ready: BABA at $109.30 is 7.1% below its 50-day average of $117.70 (that condition is met), but the Bollinger midband reclaim ($117.77), the stochastic cross (6.7 vs 31.2), and KWEB's confirmation above its $26.75 50-day average (it closed at $24.60, RSI 25.9) are all unmet. The 60-month backtest (16 trades, 56.3% win rate, 2.1% max drawdown) applies only to entries taken at the trigger, and the 12-month window produced zero trades, so recent history does not add conviction. What flips the verdict: BABA closing back above $117.77 with a stochastic cross while KWEB closes above $26.75 — or, negatively, the next annual filing showing continued margin erosion would kill the thesis entirely.

Trade now

BABA closed at $109.30 on Friday, about 7.1% below its 50-day average of $117.70, so the first entry condition — price at or below the 50-day average — is already met. The rest of the setup is not. The close must reclaim the Bollinger middle band at $117.77, which sits 8.5% above Friday's close, and the fast stochastic (currently 6.7) must cross above its slower line (currently 31.2) — a momentum turn that hasn't happened yet. Critically, the sector-confirmation leg also fails: KWEB must close above its own 50-day average, but it closed at $24.60 versus its $26.75 average, roughly 8% short. All four conditions must line up, and today only one does. So the concrete instruction is: wait. "Wait" here means watching for a single session where BABA's low touches or breaks the 50-day average, the close finishes back above $117.77, the stochastic cross prints, and KWEB closes above $26.75 — a multi-day sequence, not a one-print event. Do not pre-buy the dip; the strategy's own exit rules put a hard stop at 2.4% below entry and take profit at 4.9% above entry or at the first resistance level, so entering before the trigger changes your reward-to-risk, not just your entry price. On the flip side, the invalidation is mechanical: if BABA's close crosses back below the 50-day average after an entry, or a position falls 2.4%, the trade exits — no discretion. The 60-month backtest for this pair (16 trades, 56.3% win rate, 2.1% max drawdown) reflects entries taken at the trigger, not at arbitrary dips like today's, which is exactly why the trigger discipline matters. A nearby-parameter robustness check did not complete within its evaluation budget, so the published rule set stands as-is rather than a tuned variant.

Why the bull case still has support

The idea's core claim — that AI funding is rotating into Chinese tech rather than drying up — has at least one concrete confirmation: per Reuters on September 11, 2026, Chinese AI developer Z.AI launched $5 billion in Hong Kong share and convertible bond sales, a deal size that shows global capital still writing large checks for the theme. Yahoo Finance's September 10 report on Oracle's beat, with cloud revenue surging, supports the adjacent claim that the AI spending pipeline is real rather than hype. If the funding and spending sides of the trade are both intact, the argument goes, the laggard leg — Chinese tech — is the discounted expression. On the numbers, Alibaba is not priced like a company breaking down. Revenue for fiscal 2026 came in at $148.4B, up 8.1% year over year, with gross margin holding essentially flat at 39.8% versus 40.0% the prior year despite the heavy AI buildout narrative. Even after the operating margin hit to 4.9%, that still ranks in the 64th percentile among 854 Information Technology peers, and the 9.8% ROE sits in the 70th percentile of 682 peers. This is a company that remains mid-pack-to-good on core economics even in a down year, which is what "buying the theme at a discount" actually requires. The balance sheet gives the thesis room to be wrong without breaking. Alibaba holds $19.1B in cash against $276.8B in total assets and $153.8B in equity, with a current ratio of 1.28 and total liabilities of $113.6B — no leverage red flag in the data. Net income was $15.0B on diluted EPS of $0.80, and the per-share count fell to 1.86B from 18.47B (a 90% reduction likely reflecting an ADS/share-count rebasing rather than pure repurchase, but the direction of per-share compounding is favorable either way). The rule setup matches the thesis mechanically. Entry requires a stock to tag its 50-day average, reclaim the Bollinger midline, and trigger a stochastic cross, with the peer (KWEB for BABA entries)…

Scores

• Conviction score breakdown: 45
• Thesis support: 55
• Trade readiness: 25
• Risk quality: 55
• Backtest evidence: 55
• Fundamentals trend: 35

Watch items

• BABA — Close vs Bollinger (20) middle band
• BABA — Stochastic (14) vs Stochastic (3)
• KWEB — Close vs EMA (50)
• BABA — Close vs EMA (50)
• BABA — Insider net open-market activity (13F period 2026-09-30)
• BABA — Operating margin (FY, next annual report)
• BABA — Next dividend payment per share
• BABA — Close vs first resistance