Anthropic posted 411 open roles as of July 9, 2026, and roughly 33, or approximately 8 percent, are remote-eligible. For the company sitting at rank one on ENTRA's AI Employer Index, rated AAA across all four talent dimensions, this is not an oversight. It is a deliberate position.
The Remote Issue asks which AI employers are genuinely building distributed workforces. Anthropic's answer is nuanced enough to be worth reading carefully. It is not a remote-first company and does not claim to be. But its remote track covers a specific set of roles that no SF-only funnel would fill: senior interpretability researchers, policy scientists, trust and safety leads. Those 33 listings reach a different talent pool than the remaining 378. That gap is the story.
The Physical Bet
Anthropic has placed one of the largest physical bets in post-pandemic San Francisco. The company's SoMa footprint now spans more than 900,000 square feet across at least five buildings in the Foundry Square corridor. The cluster includes 500 Howard (its primary headquarters, approximately 240,000 sq ft), 300 Howard (466,000 sq ft, signed in early 2026), 405 Howard, 400 Howard, and 505 Howard. Total leased space crossed one million square feet in April 2026, per The Real Deal and CoStar data.
That is an infrastructure argument about research density. Anthropic's model training, interpretability tooling, alignment research, and agent infrastructure teams are physically co-located in a cluster that the company has actively expanded into while most of the tech industry continues to retreat from downtown SF. As of mid-2026, approximately 74.5 percent of Anthropic's estimated 4,000-plus employees are US-based. San Francisco anchors the majority of that figure.
Smaller teams operate in New York, London, and Dublin. The company also has in-office presence in Bangalore, Seoul, and Tokyo. None of those locations are described by the company as remote hubs. They are satellite offices, not distributed nodes.
Where Remote Actually Exists at Anthropic
The remote-eligible roles are not distributed randomly across functions. They concentrate in four areas.
Interpretability research is the most prominent. Anthropic's mechanistic interpretability team, led by Chris Olah, has posted fully remote Research Engineer listings explicitly. The team's stated philosophy, shared by Olah directly in a February 2026 posting, is that no prior interpretability experience is required and that ML infrastructure depth is the entry criterion. That framing matters. It means the remote listing is not a mid-level accommodation. It is a deliberate reach into a senior research pool that may not be in San Francisco.
Policy and Trust and Safety roles carry the next-largest share of remote listings. Anthropic's policy function needs engagement with Washington DC, Brussels, London, and Geneva. Hiring that function remotely is structurally rational. Trust and Safety roles that require cross-cultural coverage of content and model behavior are similarly well-suited to distributed hiring. These positions typically sit at the L4-L5 equivalent band, and the talent pool that holds senior policy expertise with AI fluency is geographically dispersed.
Security engineering also surfaces in the remote-eligible pool. Anthropic's ASL-3 protocols under the Responsible Scaling Policy version 3.0 create genuine internal security infrastructure requirements. Several senior security engineering roles are listed without location anchoring.
Outside those three clusters, remote eligibility at Anthropic is narrow. Engineering and research roles below the senior IC threshold (roughly L4 and below) are almost entirely SF-anchored.
Headcount Trajectory: 2021 to Now
Anthropic was founded in 2021 with approximately 120 employees, most of them research scientists and engineers who left OpenAI with Dario and Daniela Amodei. By early 2023, the headcount was approximately 240, per labor-market signals. By end-2024, it reached an estimated 1,376 (Revelio Labs). By mid-2026, the range across methodologies is 3,200 to 4,500 depending on whether extended workforce is included.
The company does not publish official headcount figures. The most reliable public proxy is the open-role count trajectory: 87 active postings in early May 2026, 392 by early June (per the prior ENTRA H1 briefing), and 411 as of July 9. That 4.7x increase in open postings over two months reflects active growth, not a plateau.
The workforce split by function is estimated at roughly 35 percent Research and AI Safety, 30 percent Engineering, 15 percent Go-to-Market and Sales, 10 percent Policy and Communications, and 10 percent Operations and Finance. The remote-eligible pool concentrates in the Research/Safety and Policy slices.
Compensation: SF Bands, Remote Reality
Anthropic does not publish official compensation bands. What the market knows comes from Levels.fyi self-reports, candidate accounts, and offer letters that have surfaced in recruiter conversations.
The picture by level: L3 (entry-level Member of Technical Staff) lands near $300,000 total compensation. L4 is approximately $475,000. L5 runs $575,000 to $725,000 for software engineers, and $700,000 to $950,000 for research scientists at the same nominal level. L6-plus software engineering roles exceed $850,000 total comp. At L6-plus, the operative variable is equity. Vesting against the $965 billion post-money Series H valuation, senior researchers on interpretability and alignment tracks have reportedly seen twelve-month realized compensation exceed $1 million in the most recent tender cycle.
There is no publicly documented remote pay differential at Anthropic. The company has not disclosed location-adjusted bands. For roles that are remote-eligible, the compensation profile appears to mirror the SF bands, per candidate accounts reviewed by ENTRA. That is consistent with the company's public posture on compensation: transparent in philosophy (mission-aligned, equity-heavy, above-market base), opaque on the exact numbers.
Policy and Trust and Safety roles, the most accessible remote track, sit in a different band. Senior policy positions at the L5 equivalent range from approximately $180,000 to $280,000 total compensation, reflecting a different labor market than the research and engineering ladder.
The Hiring Funnel: ML vs. Policy
Anthropic runs one of the harder ML screening processes among frontier labs. The technical assessment is a 90-minute online problem split into four progressively harder levels. Each level builds on the prior one and requires clean, modular code that holds under new requirements. The funnel then runs to a hiring manager screen, a three-part final round, and reference checks. The company explicitly prohibits AI assistance in all live interviews. Candidates have been removed from processes for violations.
The policy and legal track is structurally lighter on technical screening. The final round still emphasizes values alignment and mission literacy. The expectation is that a policy hire has thought carefully about AI risk, not just regulation. The initial technical bar is set for policy reasoning, not ML depth.
The average time from application to offer across all tracks is 19 days, per Glassdoor-aggregated data from 129 candidate submissions and ENTRA recruiter-side tracking. The industry benchmark for equivalent senior research roles at frontier labs runs 47 to 64 days. The mechanism is structural. Anthropic eliminated the technical phone screen for credentialed research candidates in mid-2025, moved reference checks to the front of the funnel, and schedules offer conversations within 48 hours of on-site completion.
The Funding Context
Anthropic's most recent institutional round was the Series H: $65 billion raised at a $965 billion post-money valuation, closed May 28, 2026, led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital. That followed the $30 billion Series G at a $380 billion post-money valuation, closed February 2026. The company's ARR run rate crossed $47 billion in May 2026, per reporting ahead of the Series H close.
Anthropic filed confidential IPO paperwork with the SEC on June 1, 2026. An October 2026 listing is the current analyst consensus. Finance, legal, compliance, and people operations headcount will scale materially in H2 2026 as public-company readiness becomes the operational priority. Several of those roles (regulatory counsel, people operations, compliance engineering) are likely to carry remote eligibility that the current research-heavy job board does not yet reflect.
ENTRA Talent Index
ENTRA Talent Index: 98/100 (AAA)
Component breakdown: Hiring Velocity 96, Compensation 99, Retention 97, Mission Alignment 100.
The 80 percent two-year retention rate is the single most distinctive data point in Anthropic's talent profile. SignalFire's 2025 State of Talent report placed Anthropic first among frontier labs, ahead of Google DeepMind at 78 percent and OpenAI at 67 percent. Engineers at OpenAI are eight times more likely to move to Anthropic than the reverse. When Meta ran a talent offensive in mid-2025 with packages approaching $100 million for top researchers, Anthropic lost two people. Other labs lost dozens.
The remote strategy fits inside that retention logic. A company that has built a mission-alignment filter into every hiring track does not need to offer location flexibility as a retention instrument. It offers it selectively, to reach specific talent profiles that would not otherwise enter the funnel. Interpretability researchers who want to be in Edinburgh. Policy scientists based in Brussels. Trust and Safety leads covering Southeast Asia who cannot relocate to SF.
Eight percent of 411 roles. Thirty-three people, roughly. All of them chosen with precision.
Open-role count (411, July 9, 2026) sourced from Jobera.com employer tracking. Headcount estimates (3,200–4,500) reflect Revelio Labs LinkedIn signal aggregation and Fortune December 2025 reporting; Anthropic does not publish official figures. Series H details ($65B raised, $965B post-money, May 28, 2026) sourced from Anthropic press release and TechCrunch reporting. Series G details ($30B, $380B post-money, February 2026) sourced from Anthropic press release and Yahoo Finance. ARR run rate ($47B, May 2026) cited in Series H pre-close reporting. Compensation figures reflect Levels.fyi self-reports and candidate accounts reviewed by ENTRA; Anthropic has not published official bands. Remote-eligible role count (~33, ~8%) reflects ENTRA tracking of Anthropic Greenhouse job board, July 9, 2026; individual role eligibility subject to change. SF footprint data (900,000+ sq ft, five buildings) sourced from The Real Deal and CoStar April 2026 reporting. Retention rate (80%) sourced from SignalFire 2025 State of Talent report. Time-to-offer (19 days) sourced from Glassdoor aggregated data and ENTRA recruiter-side tracking. ENTRA Talent Index score (98/100) reflects ENTRA composite methodology applied to H1 2026 data.
Global Career PlatformFind AI talent. Find your next role.
Booking is hotels. · Airbnb is apartments. · ENTRA is global careers.