Cloudflare

Cloudflare in 2026 is the most ambitious internet infrastructure bet in the world. CEO Matthew Prince shifted strategy radically in 2024-2025: from CDN/security vendor to edge AI compute layer. Workers AI (inference across 320 edge PoPs in 100+ countries), R2 (object storage with no egress fees), D1 (distributed SQLite), Durable Objects (stateful compute). All aim to beat AWS, Azure and GCP not on training, but on inference.

Q1 2026 numbers are strong: revenue $584M (+33.5% YoY), customers with ARR >$100k at 3,840 (+30%), $500k+ customers at 460 (+47%). But operating losses: GAAP op margin -9.7%, profit margin -3.7%, operating cash flow still negative. FCF positive (+$755M) primarily through SBC add-back.

Valuation: $200 per share, market cap $70B, P/S 30.3, forward P/E 131.4 (meaningless with losses). Stock -23% from the 52w high ($260), but still the most expensive profitable... err, the most expensive software business of its size without clear GAAP profitability.

The bet: if Cloudflare establishes Workers AI as the standard for globally latency-sensitive inference (edge personalization, real-time LLM routing, chatbot hosting), that is a $50-80B TAM by 2030. If not, Cloudflare is a luxuriously valued security/CDN company pressured by Akamai, Fastly and hyperscaler bundling.

Cloudflare is a smart-money battlefield in 2025-2026. Long side: Sequoia Heritage holds 18.4M shares (pre-IPO + adds), Whale Rock top-5 position at ~$3.2B long, Tiger Global rebuilt to 2.1M shares in Q4 2025 after the 2022 reduction. ARK Invest holds NET as top-3 in the ARKK ETF (4.1% weight).

Bear side: Bridgewater Associates significantly reduced in 2025, Citadel had a small short position in Q2 2025. In November 2025, Spruce Point published a 35-page short report — criticized customer concentration, FCF quality, R2 margin profile. Stock fell 8% in one day on the publication but has since recovered.

Founder Matthew Prince (CEO) holds 11.2% of shares ($8B personal stake), co-founder Michelle Zatlyn (President) 8.1% ($5.7B), co-founder Lee Holloway 4.3% — founder stack of 23.6% in aggregate, unusually high for a public software business. Prince has never sold open-market shares, only 10b5-1 vesting vehicles.

Cloudflare at $199.78 has no traditional valuation anchor. P/E negative, forward P/E 131.4 meaningless. P/S 30.3, EV/Sales 29.1, FCF multiple 93 (market cap / FCF). The only semi-meaningful metrics are rule-of-software-quality (RSQ) and peer EV/ARR multiple.

Three models: (1) DCF at 28% revenue CAGR through 2030, terminal FCF margin 30%, WACC 9.5% → fair value $155-185. (2) ARR multiple peers: Cloudflare 27x ARR — Datadog 18.5x, CrowdStrike 26x, ServiceNow 17x. Cloudflare fair ARR multiple 20-23x → $145-170. (3) Sum-of-the-parts bull case: core network (CDN+security) as a profitable $1.5B ARR business at 8x = $12B, Workers/R2/D1 as $0.5B ARR hyper-growth at 35x = $17B, AI optionality $20B = $49B market cap = $138/share. The stock is only worth $138 without AI hyper-scale.

Average ~$145-175. Current price $200. Stock is overvalued at consensus models, fair at bull case. Bull (Workers AI to $1.5B ARR by 2028 + R2-S3 substitution): $260+. Bear (Spruce Point theses confirmed + growth slowing to 22%): $115-130.

1. July 31, 2026: Q2 2026 earnings — Workers AI ARR update (consensus $145M annualized), R2 customer disclosure, possible customer-concentration response to the Spruce Point report
2. September 2026: Cloudflare Birthday Week conference — historically the most important product announcement event (2024: Workers AI, 2025: Hyperdrive), average stock reaction +3% to +7% in the 5 days after
3. Q4 2026: Possible second short report (Hindenburg signaled interest in December 2025), or resolution of Spruce Point issues via a detailed Q3 investor day

Cloudflare is the toughest valuation story in my portfolio coverage. Operationally a top software company with an impressive founder team, product vision and growth. But at $200 you pay for a perfect Workers AI scaling that has not been proven yet.

Currently not in the portfolio. Would build a 3% position below $145 (-27% from price, midpoint of models), scale to 6% at $115. Even at that low, P/S would still be 17-18 — so not a bargain, but justified for edge-AI optionality.

I take the Spruce Point concerns seriously — the customer concentration question must be answered transparently in 2026, otherwise the short-squeeze/multi-reset risk becomes a self-fulfilling prophecy. A clean Q2 2026 with top-50 customer concentration disclosure could change my bearish view. This is not investment advice — edge-AI bets are binary.