Google DeepMind Salary (2026)

DeepMind Sits on Google Leveling

After the April 2023 merger of Google Brain and DeepMind, the combined org adopted Google's standard L3 to L8 leveling. Research scientists and research engineers map to the same numeric levels as software engineers, with a parallel track designation that affects bonus and grant structure but not base. L3 is entry level for fresh PhDs in research, L4 is mid, L5 is senior, L6 is staff, L7 is senior staff, L8 is principal.

Levels.fyi data for Google AI roles places L4 total comp at $280K to $360K, L5 at $475K to $625K, L6 at $725K to $950K, and L7 at $950K to $1.4M. These bands sit below OpenAI and Anthropic at the top tiers but above most of FAANG outside the AI orgs. The Gemini team specifically has been reported to operate inside elevated bands, with research scientist offers running 10 to 20 percent above the standard Google research scientist rate.

Bay Area vs London Compensation

DeepMind's historic headquarters is London, and roughly half of the research org still sits there. London compensation is denominated in GBP and runs materially below the Mountain View band when converted to USD. A London L5 research scientist typically lands at £210K to £290K total comp, which translates to roughly $265K to $365K. The same level in Mountain View earns $475K to $625K.

The gap is partly cost-of-living, partly tax. London base salary keeps more purchasing power than the headline number suggests once Bay Area housing and California state tax are factored in. Still, retention is harder in London: researchers regularly transfer to Mountain View or accept offers from US-based labs to close the gap. Google has responded with London-specific equity refreshers for retained senior researchers but has not closed the base differential.

Equity Structure and Vesting

Google uses standard public-company RSUs on a four-year vest with the front-loaded schedule introduced in 2022 (33 percent year one, 33 percent year two, 22 percent year three, 12 percent year four). This is different from OpenAI's PPUs or Anthropic's tender-priced private equity: Google RSUs vest into liquid GOOG shares that can be sold at market on the next vesting date.

The trade-off is predictability versus upside. A four-year DeepMind L5 grant priced at $1.6M today will be worth roughly $1.6M plus or minus stock movement when it vests. An OpenAI L5 grant priced at $2.4M today could be worth $5M after the next tender. Researchers who value financial predictability and want to avoid concentration risk in a private company often choose DeepMind for this reason. Researchers who want the highest possible expected value, even at higher variance, choose OpenAI or Anthropic.

Retention Pressure from OpenAI and Anthropic

Demis Hassabis has spoken publicly about the talent retention challenge. DeepMind has lost senior researchers to OpenAI and Anthropic at a steady rate since 2023, with the most prominent departures involving researchers who built early Gemini infrastructure. Google's response has been targeted: large equity refreshers for specific named researchers, accelerated promotions, and the post-merger consolidation that put DeepMind leadership in charge of all Google AI research.

The structural disadvantage is real but bounded. DeepMind cannot match OpenAI's L5 $1.15M datapoint with a standard offer. It can match the realized number for a key researcher with a custom retention package that combines L6 promotion, a special equity grant, and a multi-year commitment. The retained researcher gets Google's brand, longer time horizon, and liquid stock; OpenAI offers higher absolute upside if the valuation keeps climbing.

Research Scientist vs Research Engineer vs SWE

The three tracks at DeepMind earn similar money at the same level, but the path to each is different. Research scientists almost always have a PhD and a publication record. Research engineers can come from either an academic or industry background and focus on the infrastructure that makes research possible. Software engineers handle product, serving, and platform work.

Base salary is identical across tracks at a given level. Equity grants are typically 5 to 15 percent higher for research scientist roles at the same level, reflecting the smaller talent pool. Research engineers fall in between. At L7 and L8, the differences narrow because individual retention packages dominate the standard band math.

Related Salary Guides

• AI Company Salaries Hub
• OpenAI Salary (2026)
• Anthropic Salary (2026)
• Meta Superintelligence Labs Salary (2026)
• AI Lab Levels Explained
• AI Lab Millionaires