Whale Rock Capital Management
Alexander (Alex) Sacerdote (founder, CIO)
Tech/TMT specialist that has leaned hard into the AI hardware supply chain — storage, PCBs, optical, and electronics-manufacturing names that feed the data-center buildout.
- Not open — Not an open-ended vehicle — analysis only, not an available allocation.
- Self-reported — Headline returns are manager- or firm-reported and unaudited.
- Partial 13F — The disclosed book is longs-only and ~45 days lagged — a slice, not the strategy.
Top holdings
Top position
AI-storage/NAND; top Q1 2026 buy (+~$270M)
AI server / electronics manufacturing
AWS hyperscaler
PCBs for AI hardware
Optical interconnect for AI networking
Recent moves
Rotated into the AI-hardware mid-cap supply chain — SanDisk, Celestica, TTM, Fabrinet, Advanced Energy, MKS, Corning — while trimming Nvidia and adding Apple/Microsoft/AMD; top sell was AppLovin.
Loading disclosed positions…
Our take
Whale Rock is making a 'second-derivative' AI bet — owning the storage, optical, PCB and components suppliers that get pulled along by hyperscaler capex, which can offer more torque (and more volatility) than owning NVDA directly.
S-curve investing is a real framework and Whale Rock applied it early — the supply-chain rotation was on before the mid-caps re-rated, and H1 2026 vindicated the call at a scale few managers reach.
Our skepticism moves rather than disappears. Three things separate out. First, an +850% move in the largest supply-chain position means the book is now far more concentrated than the March 13F shows: the win itself has become the risk, and no public filing yet reveals whether it was trimmed. Second, the framework tells you when adoption accelerates, not what a business is worth once it does — nothing in S-curve logic supplies an exit rule. Third, SanDisk, SK Hynix, Celestica and TTM remain cyclical, capital-intensive price-takers selling to a handful of buyers with enormous negotiating leverage. The position underwrites a structural re-rating of component economics. That case is stronger than a year ago but still unproven — these are one factor expressed four times, with an unobservable private Anthropic mark layered on top.
- 2026-08Aug 1, 2026: +72.5% is a Jun 30 mark, not a current one. July's hardware rout hit the four names carrying the book, and no July fund return has been published — the verdict is reopened rather than settled.
Thesis
Whale Rock is a single-factor AI-capex fund in the costume of a diversified tech book — and in H1 2026 that factor paid spectacularly. The Q1 2026 13F ($7.74B long US equity, 35 names, filed 15 May 2026) put ~16.5% in three mid-cap hyperscaler suppliers (SanDisk 6.6%, Celestica 5.5%, TTM 4.4%), plus Fabrinet, alongside GOOGL 9.9% and AMZN 4.9%. The flagship then returned +72.5% through June 2026 on SanDisk (+850%), SK Hynix (+300%) and TTM (+170%). The 'second-derivative' framing implies diversification away from Nvidia; mechanically it is still one capex bet at higher beta.
Founded 2006 by Alex Sacerdote (ex-Fidelity tech PM). Stated method is S-curve adoption analysis: own the companies positioned to dominate the steep middle of an adoption curve. Concentrated, long-biased tech, run across a flagship hedge fund, a long-only vehicle, a hybrid public+private fund, and a MegaCap Tech fund launched in 2026 (~$1B, 8–15 mega-cap names). Current expression: trim Nvidia, own the storage, optical, PCB and EMS names pulled along by hyperscaler capex.
- 2026-07The +850% SanDisk figure is a first-half mark. SNDK fell roughly 45-55% in July 2026 — the S&P 500's worst decliner that month — so the live gain is near half the quoted level.
Assessment
- S-curve inflection framing is a genuine analytical lens, and the supply-chain rotation was early rather than a crowded-trade follow-on — H1 2026 is direct evidence of that.
- Willingness to trim the consensus winner (Nvidia) while staying in the theme shows the process is not simply momentum-chasing.
- Deep domain grounding: Sacerdote's semis/hardware background at Fidelity maps directly onto the components and EMS names driving the book.
- Concentration is honest — the top six were ~35% of the disclosed long book — rather than a diversified-looking book quietly carrying one factor.
- Four of the top six are the same trade: hyperscaler capex. Correlation converges toward 1 in a capex-digestion scare, so the stated diversification is largely cosmetic.
- The book requires component/EMS margin expansion to be structural, not cyclical. H1 2026 price action is not proof of that — it is the market pricing the possibility.
- After an +850% move in SanDisk, position sizing is the open question. Public filings lag by a quarter, so nothing discloses whether the win was harvested or left to run.
- Key-person concentration is near-total. Sacerdote founded the firm in 2006 and is the franchise; no publicly disclosed succession structure.
- Valuation-agnostic frameworks give no exit discipline, and AUM growth to ~$19B plus a 2026 fund launch adds capacity strain at a point of maximum enthusiasm.
- 2026-07July 2026 inverted the trade: NVDA +0.3% and AVGO +3.1% held up while the supplier book fell — SanDisk about -47%, SK Hynix -28%, SOX -20.6%. Trimming Nvidia to own the suppliers cost the month.
- 2026-07Jul 10, 2026: the capacity concern has a dated instance — Whale Rock was a cornerstone investor in SK Hynix's Nasdaq ADR listing (priced $149, debut close $168.01); the shares then fell about 28% over the month.
Record
Attribution must be read by vehicle, since the classes diverge sharply. Hybrid fund: −9% (2021), −45% (2022), +18% (2023) — compounding to roughly −41% over three years. The +31.8% and +59.3% frequently quoted for 2023 are the public-securities class and the long-only fund, not the hybrid; conflating them overstates the recovery. The flagship was reported +51% through November 2024, then −20% in Q1 2025 before finishing the first nine months of 2025 at +14% (long-only +36%). 2026 is the outlier: flagship +72.5% and long-only +82% through June, driven by SanDisk, SK Hynix and TTM, plus a private Anthropic stake marked against a reported $965B round. Firm AUM ~$19B (July 2026). Separately, 13F-aggregator '1-year return' figures are simulated from disclosed longs — they ignore shorts, hedges, timing and fees, and are not fund returns.
Risks & fit
- Hyperscaler capex guidance flattening or being pushed out — the single shared driver behind the largest cluster in the book.
- Memory/NAND pricing is notoriously cyclical, and after an +850% run the largest position now carries far more downside convexity than upside.
- Customer concentration at the suppliers: a handful of buyers can compress EMS and PCB margins at contract renewal.
- Liquidity mismatch — mid-cap positions of this size are slow to exit if the theme breaks, which amplifies drawdowns.
- The hybrid vehicle's private holdings, including the Anthropic mark, are invisible in any filing, so both exposure and mark uncertainty are unobservable.
The 'durable edge' reading is supported if the supply-chain names hold expanded gross margins through a full capex digestion cycle — revenue flat or down, margins intact — and if Whale Rock is shown to have cut exposure ahead of a capex deceleration rather than after it. The 'levered capex beta' reading is confirmed instead if the next capex scare produces another 20%+ quarterly drawdown with the suppliers falling together (Q1 2025's −20% is the precedent), or if a full-cycle 2021–2026 compounded return for one named vehicle lands near passive large-cap tech despite the 2026 spike.
Analytical context for readers studying how tech-specialist funds express the AI-infrastructure theme, and how to attribute a record across share classes and between beta, recovery arithmetic and genuine selection. It is not an assessment of whether anyone should allocate capital. Whale Rock is a private fund with restricted access, and capacity has opened and closed at the manager's discretion. Nothing here is a recommendation to buy, sell, or mirror any disclosed position.
Fee terms are not publicly disclosed and we will not assume a 2/20 structure. The arithmetic point stands regardless: a performance-fee load on up-years within a hybrid series that compounded negatively from 2021 to 2023 is a material investor drag.