Sundheim's D1 Capital bought several tech stocks last quarter — with one big exception

D1 Capital fully exited its Meta position, selling more than 376,000 shares worth over $240 million, while also zeroing out Synopsys and Arista Networks. Sundheim increased Amazon by more than 34% to a position worth about $376.5 million and added to AI holdings in Broadcom and Nvidia, while opening new stakes in Alphabet, ASML and Taiwan Semiconductor. Instacart remained the fund's largest holding at $845 million.

Analysis

D1’s quarter-end repositioning looks less like a broad de-risking from tech and more like a deliberate rotation away from crowded, execution-sensitive software/internet names into higher-beta AI infrastructure and platform winners with clearer operating leverage. The meaningful cut in META, SNPS, ANET, and SPOT suggests Sundheim is avoiding names where multiple expansion now requires flawless ad/consumption trends, while the additions to AMZN, AVGO, and NVDA imply a preference for beneficiaries of capex intensity and model deployment rather than pure narrative exposure. That matters because the current market is rewarding the picks-and-shovels layer of AI more consistently than consumer-facing software.

The second-order read-through is that cloud and AI capex remains the key spend vector, which helps semis, networking, and industrial-capex-adjacent suppliers, but it also raises the bar for software monetization. If enterprise AI spend keeps shifting from pilots to production, the winners are the firms selling compute, bandwidth, and deployment tooling; if it stalls, those same names can compress quickly because expectations are already elevated. META’s sell-down is notable as a signal that ad-driven platforms may be entering a period where upside has to come from efficiency and AI-enhanced targeting rather than multiple recovery alone.

The most actionable setup is a relative-value expression: long AMZN/AVGO/NVDA versus short META/ANET/SNPS, because the former cluster has a cleaner path to near-term estimate revisions from spending and product cycle tailwinds, while the latter faces longer-duration proof points. The risk is that the market has already partially anticipated this rotation, so chasing the long side after the rebound could reduce upside if AI capex decelerates in the next 1-2 quarters. Contrarianly, the big AMZN increase may be less a conviction call on e-commerce and more a hidden bet on margin expansion from ads, logistics, and cloud mix, which could still be underappreciated if consumers remain stable and AI spend stays elevated.

Key Decisions for Investors

• Go long AMZN vs. short META for a 2-3 month window: target a 10-15% relative move if ad/commerce efficiency and cloud leverage keep driving upward revisions; stop if META re-accelerates ad growth or AMZN margin guidance disappoints.
• Pair long NVDA or AVGO against short SNPS or ANET: own the names with direct AI capex pass-through and shorter earnings translation, while fading the names that need sustained design-win conversion to rerate.
• Add to AMZN on pullbacks only, not strength: use intraday or 3-5% retracements to limit entry risk, because the stock has already moved sharply and near-term reward/risk is better on dips than breakouts.
• For a cleaner hedge, short a basket of META/SNPS/ANET into earnings season and take profits into any 5-7% drawdown; these names are more vulnerable to guidance resets than to valuation support.
• Watch for a 1Q/2Q capex deceleration signal from hyperscalers: if it appears, reduce AVGO/NVDA exposure quickly, as the trade thesis relies on sustained infrastructure spend rather than one-quarter positioning noise.