Remote tech jobs in 2026: one in three postings is still fully remote

Remote tech jobs in 2026 are not disappearing. Across 20,350 tech jobs posted to Greenhouse, Lever, Ashby, and Rippling between June 26 and August 25, 2026, fully remote roles made up 33.9% of new listings. Hybrid roles made up 33.9%. Onsite roles made up 32.2%. The three work modes split the market in thirds.

That contradicts the dominant narrative. Robert Half's Q2 2026 analysis of U.S. job postings from TalentNeuron puts technology at 85% fully on-site, 11% hybrid, and 4% fully remote. JobScroller's 2026 report, built from 106,000 scraped listings, puts 85% of tech roles in-office and 14% fully remote. Both numbers are widely cited as proof that remote work is over.

Our number comes from a different source: the work-type field on the applicant tracking systems themselves, read on every re-crawl at 900 companies Hotfix tracks continuously. The return-to-office wave is real at some employers. It is not the whole market.

Remote tech jobs in 2026: the headline split

Of 20,350 tech jobs posted in the 60-day window ending August 25, 2026:

• Fully remote: 6,906 listings (33.9%)
• Hybrid: 6,894 listings (33.9%)
• Onsite: 6,551 listings (32.2%)

Hybrid briefly passed remote in July, when hybrid hit 35.0% and remote fell to 33.1%. By August the two had converged again at 35.0% remote and 32.5% hybrid. Onsite recovered from a February low of 24.4% to 32.5% in August. None of the three modes approached the 85% on-site share that aggregate labor-market reports describe for tech.

What we measured

Hotfix ingests jobs directly from applicant tracking systems: Greenhouse, Lever, Ashby, and Rippling. This analysis covers every job posted between June 26 and August 25, 2026, at the 900 companies in our index. Work type comes from the ATS classification on each listing (remote, hybrid, or onsite), not from parsing the location field or inferring arrangement from the job title.

That distinction matters. JobScroller classifies work mode from the location text and counts any listing with a city but no remote label as in-office, which its own methodology notes likely undercounts hybrid. Robert Half uses TalentNeuron data mapped across 450+ job titles with an LLM-based classifier. Both approaches cast a wider net across the whole economy. Ours reads what the employer told the ATS at the moment of posting.

Company size splits the market

Return-to-office is concentrated at large employers, not uniform across them.

At companies with 1,000 or more employees, hybrid leads: 40% of postings are hybrid, 32% onsite, and 28% fully remote. At companies with 201 to 1,000 employees, remote leads: 49% remote, 30% hybrid, and 22% onsite.

The bifurcation has names on both sides.

Remote-first large employers still post almost entirely remote roles: GitLab (100% remote across 264 listings), Twilio (100%), Dropbox (100%), and Instacart (94% remote). Coinbase posts 94% remote across 208 listings.

Onsite-first large employers post the opposite profile: Crusoe (99.6% onsite), Applied Intuition (99.5% onsite), xAI (95.5% onsite), and Flexport (94.2% onsite).

Amazon's five-day return-to-office mandate, effective January 2025, is the headline that gets repeated. It is one employer's policy, not a market-wide collapse of remote hiring.

Engineering roles stay the most flexible

Technical roles remain the segment where remote and hybrid options are widest.

Among jobs posted in the last 60 days, data engineering leads at 40.5% remote and 33.2% hybrid. Backend engineering follows at 39.2% remote. Solutions architecture sits at 37.7% remote and 34.4% hybrid. Machine learning engineering is the outlier among technical functions: 25.5% remote and 36.3% hybrid, consistent with frontier AI labs clustering around in-person work.

For job seekers filtering on remote software engineer roles, the underlying market still supports that search. The constraint is competition, not absence.

Entry-level remote is tighter

Remote share rises with seniority up to the senior level, then dips slightly at staff.

Entry-level candidates face a thinner remote market. That compounds a problem we measured in June: only 1 in 10 tech engineering jobs is open to entry-level candidates. Dropping the remote filter removes roughly one in four entry-level openings. It does not fix the seniority bottleneck.

Remote availability and remote pay are separate questions. We covered pay in July: remote engineering jobs advertise 13% below onsite across the market, but the gap inside a single employer is 0.3%. The remote filter sorts candidates into a different employer pool more than it prices the commute.

Why the published numbers disagree

Three factors explain the gap between our one-third remote finding and the 4% to 14% figures in aggregate reports.

Dataset scope. Robert Half and JobScroller scrape broad job-board and aggregator listings across the economy. Hotfix reads supported ATS feeds at companies that publish through Greenhouse, Lever, Ashby, or Rippling. Tech-native employers that hire through those systems are overrepresented relative to the full U.S. labor market.

Classification method. Inferring work mode from a location string treats a San Francisco listing with no remote label as in-office even when the employer intended hybrid. Reading the ATS work-type field avoids that error but misses employers who post only on LinkedIn or their own careers page.

Worker-side vs posting-side. WFH Research tracks where remote-capable workers actually work, not what employers post. Its 2026 data still shows roughly 60% of remote-capable tech workers in remote or hybrid arrangements. Posting mix and worker mix measure different things. Both can be true.

What this means for job seekers

Remote tech jobs in 2026 exist at scale. They are not the default arrangement they were in 2021, and they are not the 4% sliver that headline statistics imply. The market settled into a three-way split: remote, hybrid, and onsite each claim roughly a third of new ATS postings.

The practical filter is employer, not work mode alone. Remote-first companies still post hundreds of roles. Onsite-mandated companies post hundreds more. The split shows up on remote software engineer roles and among roles with disclosed salary ranges, not in aggregate RTO headlines.

What this analysis cannot show

This dataset covers employers on supported ATS platforms. It does not include every company in the U.S. economy, and it underweights employers who post exclusively on LinkedIn, Indeed, or proprietary career sites. A warehouse operations role at a non-tech employer will not appear here.

We also measure what employers post, not what they enforce after hire. A listing marked hybrid may become five-day onsite in practice. Our finding is about advertised arrangement at the time of posting.

WFH Research surveys show 39% of job seekers rank remote flexibility as a primary reason to change jobs in the second half of 2026 (Robert Half). The postings are still there. The competition for them is the part that got harder.