We're booking big profits in an AI stock and trimming a credit-exposed bank

Jim Cramer's Charitable Trust is selling 40 Broadcom shares at about $335, lowering the position to 120 shares and its portfolio weight to 1% from 1.36%; the sale realizes a 312% gain on shares bought in August 2023. The Trust is also selling 75 Capital One shares at about $201 ahead of earnings, cutting its stake to 600 shares and portfolio weight to about 3.1% from 3.45%, and downgrading the stock to a hold-equivalent rating. The Broadcom trim reflects downside protection amid rising capital costs for AI stocks; the Capital One reduction reflects concern that investment in its Discover and Brex acquisitions has limited outsized profit beats.

Analysis

The trims are more informative about portfolio risk limits than about new company-specific evidence. For Broadcom, the relevant transmission channel is valuation duration: higher capital costs can pressure AI-related multiples before they materially change demand. A sustained pullback in hyperscaler capex or weaker networking/accelerator demand would make that risk fundamental; absent such confirmation, this is a positioning signal, not an AI-demand call. Competitors and suppliers may move with the same factor, so a broad semiconductor hedge could dilute the intended exposure reduction.

Capital One’s risk is more idiosyncratic: acquisition integration and investment spending may delay earnings contribution, while consumer credit deterioration or funding pressure could compound the drag. Bank of New York Mellon offers different exposure, but it is not a clean hedge; its earnings remain sensitive to rates, markets, and client activity. The article provides no credit, synergy, or valuation data to establish relative value.

Near term, COF’s earnings can reset expectations and produce a gap; avoid adding ahead of the release without examining credit trends, expense trajectory, and integration guidance. Over 1–3 months, AVGO’s key checks are customer capex commentary and AI-related revenue/guidance. Over 6–18 months, successful acquisition integration or durable AI infrastructure spending could reverse the cautious positioning. The contrarian point: trimming after a rebound can be prudent risk control, but extrapolating one investor’s exits—or a higher cost of capital alone—into a sector top is not supported here.

Key Decisions for Investors

• AVGO: Treat the trim as a modest de-risking cue, not a standalone short signal. Keep exposure sized to tolerate multiple compression; reassess if hyperscaler capex commentary or Broadcom’s AI-related guidance weakens.
• COF: Wait for earnings before changing exposure. Monitor delinquency/charge-off trends, funding costs, expense guidance, and quantified acquisition synergies; worsening credit metrics alongside continued investment would strengthen the downside case.
• BNY versus COF: Do not assume BNY is a direct defensive substitute. Consider a relative-value position only after comparing valuation, deposit/funding sensitivity, fee revenue, and credit exposure; those data are not supplied here.
• Falsification/watch list: A sustained recovery in AI customer spending and AVGO guidance would weaken the capital-cost thesis; improving COF credit indicators and clearer integration payback would challenge the cautious COF view.