GOOGL options restrategy

1 · Your question: the Jan-2028 $330 short put

Sell the 2028-01-21 $330 put at the bid ($43.00/sh, mid $44.70; OI 333, thin - work the mid):

• In-the-money boundary: $330. GOOGL only has to fall 1.7% for it to expire ITM - it's barely below spot, so expect it to be ITM along the way frequently.
• Profit boundary at expiry: $330 - $43 = $287, i.e. -14.5% from here. Anywhere above that at expiry you make money; below it you lose ~$100 per $1.
• Max profit: $4,300/contract if GOOGL ≥ $330 in Jan 2028 - that's 13.0% on $33k cash-secured over 17.7 months (~8.8%/yr), more on margin.
• Probabilities: your saved distribution puts P(below $287 at Jan-28) at 33%; the market-implied (risk-neutral, skew-fattened) says 43%.

2 · Position triage: what the calls are worth now

No $545 strike exists at Jan-2028 (strikes: 520/540/560/580) - confirm from your broker statement which you own. Under your saved distribution P(GOOGL > 540 at Jan-28) ≈ 11%.

3 · The redeploy: Jan-2028 expiry scenarios (per contract)

Put spread = sell 330P at $43.00 bid, buy 230P at $11.55 ask (your "emergency brake ~$100 lower"). It caps max loss at $68.55/sh vs $287/sh theoretical for the naked put, costs ~$1,155 of EV in the middle band, and roughly zeroes the edge under your saved view (EV ≈ +$4 vs +$371 naked). What it really buys: crash-path survivability and less margin drawdown.

4 · Collaring the 1,114 shares (~$374k)

11 contracts covers 1,100 of your 1,114 shares. Interactive version with per-band probabilities: report §3 (hedge builder, put-spread-collar variant in §3c).

5 · The DeepSeek precedent (Jan 27, 2025) vs now

• Kimi K3 (Jul 16 & 27, 2026) is NOT the main driver this time. It caused a ~-1.5% Nasdaq day and GOOGL -1.5% premarket - a flashback, not a shock. Morningstar et al. called the fears misplaced.
• The real hit was Alphabet's own print (Jul 22): revenue +24% and Cloud +82% beat, but capex guided up to $195-205B, first negative-FCF quarter (-$5.9B) since the 2004 IPO, and 2027 capex to "significantly increase." Stock -6 to -7.4% next day, worst in a year.
• Your "pullback at 2027 planning season" thesis is so far unconfirmed: the current signal is escalating spend (analysts float >$1T hyperscaler capex in 2027) plus rising investor scrutiny - sentiment flipped from "more capex = good" to "show me payback." No hyperscaler has guided capex down. MSFT + META report Jul 29 (today), AMZN Jul 30 - the live test of the thesis is this week.

6 · CapEx composition: volume or cost inflation?

Verdict: both, and Alphabet won't say the split. Alphabet frames its $195-205B guide-up purely as volume ("acceleration in the delivery of capacity to meet growing demand"; CFO: "we're still in a supply-constrained environment"). But the two peers that did quantify it attribute a large chunk to component-cost inflation, mostly memory: Microsoft explicitly pinned ~$25B of its ~$190B 2026 capex on higher component pricing; Meta said "most of" its $10B guide-up was rising memory costs. Memory prices roughly doubled in 2026, so even a pure "build more" plan costs materially more per unit.

Why it matters for the trade: if a meaningful slice of the capex shock is memory-price inflation, the "2027 pullback" could arrive as deflation in unit costs (memory cycle rolling over) rather than hyperscalers cutting unit volume - which would let capex guides flatten while compute delivered keeps growing. That's a more benign path for GOOGL than a demand-led spend cut, and it's consistent with your "inference stays valuable" thesis.

6b · How much of Google Cloud demand is Anthropic?

The clean number is backlog, not revenue: Anthropic ≈ 40%+ of Google Cloud's $514B backlog; its share of current recognized revenue is much smaller and undisclosed. The April 2026 $200B/5-year commitment (TPU capacity starting 2027, per The Information) plus the Oct 2025 up-to-1M-TPU deal (~$50B est.) dominate the RPO. Growth attribution today is only qualitative ("long-term contracts from major clients such as Anthropic").

7 · Margin: does the long 230P reduce the requirement? Yes.

Yes - buying the 230P converts an open-ended margin requirement into a capped one. Brokers margin the pair as a defined-risk spread, not two separate positions (same account, same underlying, long expiry ≥ short expiry - true here, both Jan-2028):

• On IBKR Portfolio Margin, requirements are risk-based (stress-test), typically well below Reg-T for both, but the same principle holds: the naked put's requirement expands with vol and spot drawdown; the spread's is bounded by max loss. Preview the exact number pre-order with IBKR's "Check Margin" button.
• The collar legs need no margin at all: the short 370/375C is covered by your 1,114 shares; long puts just cost premium.
• Assignment: American-style, so a deep-ITM short put can be assigned early (usually only when time value ≈ 0). You'd be put 100 sh/contract at $330; the long 230P stays in place as protection.

8 · Tax mechanics (educational - confirm with your CPA)

• Selling the losing calls now: short-term capital losses (held <1yr). They offset any 2026 capital gains first, then $3k/yr of ordinary income, rest carries forward indefinitely. Replacing long calls with short puts is generally not a wash sale - a short put isn't "substantially identical" to a long call - so the losses should hold. (Edge case: selling puts so deep ITM they're economically a stock purchase can be attacked; your near-the-money 330s are not that.)
• Short-put premium is taxed only when the position closes - expiry, buy-back, or assignment - and writer gains are always short-term no matter how long the position was open. So the Jan-2028 330P's $4,300 lands as 2028 short-term gain (federal ordinary rates + NIIT; California taxes all cap gains as ordinary income anyway, if you're filing CA). Your 2026 call losses carry forward and can offset it.
• If assigned: no gain on the premium at assignment - it reduces your cost basis on the shares you're put (basis $330 - $43 = $287), and a fresh holding period starts.
• The collar on your employee shares is the tricky one. Your low-basis shares are presumably already long-term - a straddle can't take that away. A wide out-of-the-money collar (320/370) is generally not a constructive sale (that requires essentially locking the price). But a collar is a tax straddle: losses on one leg are deferred to the extent of unrecognized gain on the other leg, and the qualified-covered-call exception doesn't apply once the put is attached. Practical effect: leg losses may not be deductible in the year incurred - annoying, not catastrophic. Worth 30 minutes with a CPA before executing size.
• The 330/230 spread is also technically a straddle (offsetting positions in the same underlying) - same loss-deferral mechanics if you close one leg at a loss while the other has a gain.