The Remote Work Salary Gap: What Engineers Actually Earn by Work Model

Nearly half of all software engineering positions in 2026 still require full-time office presence. This may surprise observers who assumed the pandemic permanently reshaped the workplace, but the data is unambiguous: 49.2% of active listings are classified as onsite, making it the dominant work model by a significant margin. Remote positions account for 26.9% of the market, while hybrid arrangements make up the remaining 23.9%.

These numbers represent a stabilization rather than a trend. The rapid expansion of remote work in 2020 and 2021 was followed by a partial retraction in 2022 and 2023 as companies implemented return-to-office mandates. By 2026, the market appears to have reached an equilibrium. The companies that were going to pull people back have done so. The companies that committed to remote have generally stayed the course. And a growing middle tier has settled on hybrid as a compromise that satisfies neither camp perfectly but functions well enough in practice.

What is more interesting than the aggregate split is how unevenly it distributes across industries, company types, and geographies. As we will see, some sectors are overwhelmingly remote-friendly while others remain almost entirely office-bound. These structural differences matter far more for individual engineers than the headline percentages suggest, because the work model available to you depends less on your personal preferences and more on what you build and who you build it for.

The headline finding of this analysis defies the narrative that has dominated remote-work discussions for years. Hybrid roles pay the most. At $187,840 average and $183,825 median, hybrid positions outpay both remote ($175,079 average, $170,000 median) and onsite ($185,626 average, $175,000 median) roles. The gap between hybrid and remote is not trivial: $12,761 in average salary and $13,825 in median salary. That is a difference large enough to matter on a mortgage application, and it persists even after controlling for the distribution of seniority levels across work models.

This "hybrid premium" has a structural explanation that is more mundane than it might first appear. Companies offering hybrid roles tend to be concentrated in the upper tier of the employer market: large, well-funded technology companies, established financial institutions, and growth-stage startups that have raised significant capital. These organizations have invested in office space and want to extract value from it, but they have also learned from 2020 to 2022 that rigid full-time office mandates shrink their applicant pools. The compromise, two to three days per week in the office, allows them to maintain their physical presence while competing for talent that expects some flexibility. Because these employers tend to be well-capitalized, they can afford to pay above-market rates.

Remote roles, paradoxically, include the widest salary range in the dataset. The 25th-to-75th percentile band for remote positions stretches from roughly $105,000 to $228,000, a spread of $123,000. This reflects the extreme heterogeneity of the remote market. On one end, you have fully distributed companies like GitLab, Automattic, and a constellation of well-funded startups that pay competitive national-market salaries. On the other, you have companies that use "remote" as a keyword to attract applicants to roles that pay local-market rates, or contractors and agencies that list positions at below-market compensation. The median of $170,000 sits lower than the other models because the lower tail of the remote distribution drags it down.

Onsite positions land in the middle at $185,626 average and $175,000 median. This includes everything from FAANG-adjacent companies in the Bay Area paying $200,000+ to defense contractors and enterprise shops in lower-cost markets paying $130,000 to $150,000. The onsite median is $5,000 higher than the remote median, which inverts the finding from our salary guide that analyzed compensation at specific seniority levels. The apparent contradiction dissolves when you account for seniority mix: remote roles skew slightly more toward mid-level positions, while onsite roles have a higher proportion of senior and staff listings from large technology companies.

For individual engineers, the practical takeaway is this: the work model itself is a weaker predictor of salary than the type of company offering the role. A remote position at a well-funded Series C startup will almost certainly pay more than a hybrid role at a mid-size regional consultancy. The aggregate data shows hybrid paying more on average, but that is largely because hybrid is correlated with employer quality rather than because the hybrid arrangement itself commands a premium.

The average-versus-median divergence tells an important story about each work model's salary distribution. Hybrid roles show the smallest gap between average and median ($4,015), suggesting a tightly clustered distribution without extreme outliers pulling the average in either direction. These are consistent, well-paying positions at companies with deliberate compensation strategies.

Onsite roles show the largest average-to-median gap at $10,626. This means the onsite distribution is right-skewed: a concentration of roles in the $140,000 to $190,000 range with a long tail of highly compensated positions at elite technology companies that push the average upward. If you remove FAANG and adjacent companies from the onsite pool, the median barely moves but the average drops by roughly $8,000. The high-end onsite outliers, think principal engineers at Meta earning $350,000+ base, inflate the average beyond what a typical onsite engineer actually earns.

Remote's $5,079 average-to-median gap reflects the opposite asymmetry. The remote distribution has a long left tail: a substantial number of remote positions in the $90,000 to $130,000 range, often at companies hiring in lower-cost markets or offering contractor-like arrangements under the "remote" label. The highest-paying remote roles match or exceed hybrid compensation, but there are proportionally more lower-paying positions dragging the central tendency downward.

The industry breakdown is where the remote-work story gets genuinely interesting, because it reveals that remote availability is not a general tech-sector phenomenon. It is a property of specific industry verticals, driven by the nature of the work and the culture of the companies within them.

Crypto leads with 55.6% remote listings. This is not surprising when you understand the industry's origins. The cryptocurrency and Web3 sector was built on a philosophical commitment to decentralization that extends beyond technology to organizational structure. Many of the largest crypto companies, including exchanges like Kraken and protocol teams like those behind Solana and Polygon, were distributed from their founding. They did not adopt remote work in response to a pandemic. They were born that way. The 55.6% figure likely understates the true remote rate, as some crypto companies list positions as "hybrid" with no actual office attendance expectation.

Gaming at 41.7% represents one of the more unexpected entries near the top. Video game studios were historically among the most office-centric workplaces in tech, with culture built around war rooms, crunch periods, and in-person playtesting. The pandemic forced a reluctant experiment in remote game development, and many studios discovered that it worked better than they expected, at least for the programming side. Art, design, and QA teams have been slower to go remote, but engineering roles in gaming are now nearly as likely to be remote as those in fintech or SaaS.

The cluster of industries in the 31-34% range, including fintech, SaaS, cybersecurity, and DevTools, represents the tech-sector norm. About one in three positions in these verticals is fully remote. This figure has been remarkably stable since late 2024, suggesting that the remote share in mainstream technology companies has found its natural level. These industries employ enough engineers that their aggregate numbers drive the overall 26.9% figure upward, offsetting the lower remote rates in traditional industries.

The AI/ML figure of 25.0% deserves scrutiny because it contradicts the assumption that the most technically sophisticated industries would naturally embrace remote work. There are concrete reasons why AI companies pull people into offices. Training and fine-tuning large models often requires proximity to GPU clusters, not because engineers physically interact with hardware, but because the iteration cycles on training runs benefit from low-latency access to shared infrastructure and the ability to debug in real time with teammates. Many AI companies also impose strict security requirements on model weights and training data that make remote access architecturally harder to provision. And at a more cultural level, the AI research community values in-person collaboration, whiteboards, and informal discussion in a way that differs from, say, SaaS product engineering.

At the bottom of the ranking, defense (13.5%) and automotive (13.9%) are remote outliers for entirely different reasons. Defense work often requires security clearances, SCIFs (Sensitive Compartmented Information Facilities), and access to classified networks that physically cannot exist outside of approved facilities. An engineer working on missile guidance systems or satellite communications simply cannot do that work from a home office. Automotive is constrained by hardware: building embedded systems for vehicles, testing sensor suites, and integrating software with physical powertrain components all require presence in labs and testing facilities. These are not cultural preferences that might shift over time. They are structural requirements of the work itself.

The company-level data reveals just how binary the remote question is for many employers. At companies like Blue Light Consulting, Launch Potato, Truelogic, and Tether, remote is not an option on a dropdown menu. It is the only way they operate. These organizations have no offices. Their entire infrastructure, from onboarding to performance reviews to architecture discussions, is built around asynchronous communication and distributed teams. For engineers who want guaranteed remote work, these companies represent the most reliable bet.

At the other extreme, Anduril and LinkedIn post zero remote positions across their entire engineering organizations. Anduril's case is straightforward: the company builds autonomous defense systems, and much of the engineering work involves classified projects that require physical presence in secured facilities. LinkedIn's zero-remote figure is more noteworthy because the company operates in a sector (enterprise SaaS) where remote work is common. As a Microsoft subsidiary, LinkedIn follows the broader Microsoft policy that favors in-office and hybrid arrangements. The company's stated position is that in-person collaboration produces better products, a claim that is widely debated in the industry but clearly reflected in their hiring practices.

Bosch (2% remote) and Visa (5% remote) represent two different kinds of office attachment. Bosch is constrained by hardware: much of its engineering work involves embedded systems, sensors, and physical products that require lab access. Visa is constrained by regulation: payment processing involves compliance requirements that limit where and how code can be written and deployed. Both companies offer some remote flexibility at the margins, typically for senior architects or specialized consultants, but the vast majority of their engineering workforce operates on-premises.

The pattern that emerges is clear: a company's remote policy is a function of its industry, its security requirements, and its founding culture, roughly in that order. Companies born remote tend to stay remote. Companies in hardware, defense, or heavily regulated industries tend to stay onsite. And companies in the broad middle of the software industry tend to converge on hybrid as the default.

The temptation with data like this is to optimize narrowly. Hybrid pays the most, so target hybrid roles. That would be the wrong conclusion. The $12,000 to $14,000 gap between hybrid and remote, while meaningful, is a second-order variable in the total compensation equation. Seniority level, tech stack specialization, and industry vertical each exert more influence on your salary than whether you sit at a desk three days a week or work from your kitchen. An infrastructure engineer working remotely at a well-funded startup will almost certainly out-earn a frontend developer in a hybrid role at a mid-size company. The work model premium is real but small relative to these other factors.

That said, there are situations where the work model should be a primary consideration in career planning:

If you live outside a major tech hub, the remote market is where you capture the most value. An engineer in Raleigh, Nashville, or Portland earning $145,000 at a local employer can potentially access $165,000 to $185,000 by competing for remote roles benchmarked against the national market. The "remote discount" that appears in aggregate data is an artifact of the bimodal distribution, and the top half of that distribution pays as well or better than hybrid roles at mid-tier employers.

If you are targeting a specific industry, know the structural constraints before you start your search. An engineer who wants to work in defense should not expect remote options. An engineer interested in crypto should expect almost nothing but remote options. These are not trends that are likely to shift in the near term. They reflect the fundamental requirements of the work and the deeply embedded cultures of the companies that do it.

If you are optimizing for total lifetime earnings rather than current salary, the work model decision intersects with geography and cost of living in ways that the headline numbers miss entirely. A remote engineer earning $170,000 while living in a city with a cost-of-living index of 85 (say, Austin or Denver) has more disposable income than a hybrid engineer earning $188,000 in San Francisco with a cost-of-living index of 175. The salary gap tilts one way; the purchasing-power gap tilts the other.

One more consideration that rarely appears in compensation analyses: career velocity. The data cannot measure this directly, but there is a legitimate question about whether onsite and hybrid roles offer faster promotion paths due to increased visibility and stronger professional relationships with decision-makers. If the hybrid premium partially reflects the fact that hybrid employees advance more quickly to higher-paying seniority levels, then the gap may compound over time in ways that a snapshot analysis cannot capture. Engineers who prioritize long-term trajectory over current compensation may want to weight this factor alongside the raw salary numbers.

If there is a directional trend in this data, it points toward hybrid as the emerging default for mainstream technology companies. The logic is straightforward and has less to do with ideology than with economics. Companies that mandate five days a week in the office face a quantifiable hiring disadvantage. They are fishing in a geographically constrained talent pool while their hybrid competitors can draw from a wider radius. At the same time, companies that go fully remote sacrifice the incidental collaboration, mentorship, and culture-building that physical proximity provides, and some are beginning to see the effects in slower onboarding, reduced innovation, and difficulties maintaining organizational cohesion.

Hybrid emerges as the least-bad option for most employers. It preserves enough in-person time to maintain culture and collaboration while offering enough flexibility to remain competitive for talent. The salary data confirms this positioning: companies offering hybrid roles can afford to pay more because they tend to be stronger businesses with better-funded compensation budgets. The hybrid premium is not a bribe for coming to the office. It is a reflection of the fact that the companies choosing hybrid are, on average, the most successful companies in the market.

For the 49.2% of listings still classified as fully onsite, the question is whether that share will erode. In some industries, it will not. Defense, automotive, hardware engineering, and portions of biotech will remain overwhelmingly onsite because the work demands it. But in software-heavy industries like SaaS, e-commerce, and enterprise software, the onsite share is likely to continue declining as companies realize that the talent-market costs of a rigid office mandate outweigh the collaboration benefits. The movement is from onsite to hybrid, not from onsite to remote. Full remote appears to have plateaued as a share of the market, with growth coming primarily from the conversion of onsite roles to hybrid arrangements.

The implication for engineers is that the ability to work in a hybrid model is becoming a baseline expectation rather than a perk. Just as unlimited PTO evolved from a differentiating benefit to a standard offering, hybrid flexibility is transitioning from something companies advertise as a selling point to something candidates assume as a default. The employers that resist this shift will not disappear, but they will increasingly be competing for talent within a narrower pool, and the salary data suggests they will need to pay a premium of their own to overcome the flexibility disadvantage.