· Investment Strategies · 2 sources
Magentic Raises $18M Series A — Sequoia-Backed Agents for Factory Procurement
Felicis led Magentic’s $18M Series A a year after Sequoia’s $5.5M seed. Global 500 manufacturers run procurement agents inside Teams and ERPs — who buys and why Sequoia stayed.
By Venture Capital Tracker Editorial
Magentic raised an $18 million Series A announced September 17, 2026, led by Felicis, with Sequoia Capital and The Westly Group returning. TNW notes Sequoia led a $5.5 million seed at the July 2025 launch. Valuation and ARR were not disclosed.
the AI capex boom still runs through procurement teams stuck in Excel and aging ERPs — Magentic sells agents that place and clear work inside those systems, not another spend dashboard.
Key facts
- Company
- Magentic — London + New York; McKinsey / OpenAI alumni founders
- Round
- $18M Series A
- Date
- September 17, 2026
- Lead
- Felicis
- Returning
- Sequoia Capital, The Westly Group
- Prior
- $5.5M seed (Sequoia-led, Jul 2025 per TNW)
- Traction claims
- Global 500 base; 3 of top-10 beverage cos; >1M orders/yr at one customer; $4M savings at another; typical 2–5% savings, 60% data-quality lift (company)
- Security claims
- Zero-data-retention with major AI providers; any-cloud / regional isolation
- Use of proceeds
- Agent roadmap, procurement/supply-chain coverage, long-horizon optimization research
Who uses the product — and for what job
Users: procurement and supply-chain teams at large manufacturers — indirect and direct spend, including raw materials.
Job: AI digital workers that operate like virtual employees on Teams/email/ERP: decide buy vs build, pick suppliers, negotiate, run orders, clear invoices — humans stay in command.
CEO cites converging pressure from AI-driven physical capex, tariffs, and budgets that grow slower than workload (company cites Goldman Sachs ~$8T AI capex 2026–2031 as context — third-party projection, not Magentic revenue).
Why now
- Physical-world AI demand (fabs, data centers, factories) spikes procurement complexity while headcount budgets lag.
- Felicis partner Feyza Haskaraman’s quote: automating supply-chain action inside manufacturer systems is rare because integration is hard — that difficulty is the moat thesis.
- One year from seed to Series A with named Global 500 usage is the timing story.
Why Felicis / Sequoia — portfolio fit
Sequoia led seed and returned — classic conviction on a systems-integration-heavy industrial agent. Felicis leading Series A adds a specialist growth narrative without replacing Sequoia.
**
| Investor | Fit |
|---|---|
| Felicis | Lead check; thesis on physical-economy agents |
| Sequoia | Seed continuity + brand for enterprise sales |
| Westly | Climate/industrial adjacency from prior round |
Strategic rationale: keep Sequoia for enterprise trust, add Felicis as a hungry Series A lead that wants the industrial agent category, preserve Westly’s industrial network.
Competitive map
| Player | Lane |
|---|---|
| Coupa / SAP Ariba / Jaggaer | Incumbent procurement suites; Magentic pitches action agents on top of mess |
| Horizontal agent platforms | Weak manufacturing data/context |
| Supply-chain AI peers (e.g. ops automation) | Overlap on workflows; Magentic emphasizes manufacturer-scale direct spend |
Missing numbers
- No valuation, ARR, or customer count beyond Global 500 framing.
- Savings percentages are company-typical claims — not audited case studies in the release.
Practical takeaway
- Founders (industrial AI): Sell orders cleared and dollars saved inside the customer’s systems of record.
- Investors: Diligence ERP integration depth and security isolation — that is the product.
- Procurement leaders: Ask for production order volume and exception rates, not chat demos.
Follow Venture Capital Tracker on Google
Add VCT as a preferred source to see more of our venture-capital reporting in Google Search.
Editorial note: AI tools assisted with research, structure, or drafting. Venture Capital Tracker retains human editorial responsibility for factual accuracy, relevance, and source quality before publication.
