Which AI Company Pays the Most? We Read 2,144 Posted Salary Ranges (August 2026)
On August 29, 2026 we read every US job posting on twenty AI and robotics company boards — 2,936 of them — and pulled out the 2,144 that printed a salary range. Anthropic posts the highest numbers: a median range of $315,000–$405,000, midpoint $362,500. OpenAI is second at $250,000–$345,500, midpoint $307,500. Third is Cohere at $280,000, then xAI at $275,000.
That is the answer to the question as people type it. The more useful answer takes another three minutes, because two things we found while counting change what the ranking means — and one of them changes what you should do about it.
The first: the top two companies do not label their numbers the same way. The second: the company that pays the most had zero entry-level openings out of 443 US postings on the day we looked.
How we counted
We pulled every posting from the public job board of twenty AI and robotics employers on August 29, 2026 — 3,716 roles worldwide, 2,936 of them in the United States — using each company’s own Greenhouse or Ashby feed. “Published a range” means an annual dollar figure appears in the posting. For each company we took the median of the low ends and the median of the high ends, and the midpoint of those two, so that one $850,000 outlier cannot move the result. Every figure here comes from the employer’s own page on that date. We are not recruiters, we take no money from any employer named here, and postings close without notice — the official page always wins. Full method on How We Count, live totals on our AI Jobs Tracker.
Jump to what you need
- The ranking, by the numbers employers printed
- Does it survive a like-for-like check?
- The top two numbers are not the same kind of number
- The best-paying company has no way in
- A quarter of these jobs never say
- What the reachable jobs actually pay
- What moved this week
- What to do with this
- Questions people actually search
The ranking, by the numbers employers printed
Fifteen of the twenty companies published enough US ranges to rank. Here they are, biggest number first.
| # | Company | US ranges read | Median posted range | Midpoint |
|---|---|---|---|---|
| 1 | Anthropic | 427 | $315,000 – $405,000 | $362,500 |
| 2 | OpenAI | 612 | $250,000 – $345,500 | $307,500 |
| 3 | Cohere | 31 | $220,000 – $320,000 | $280,000 |
| 4 | xAI | 73 | $180,000 – $400,000 | $275,000 |
| 5 | Sierra | 107 | $200,000 – $290,000 | $262,500 |
| 6 | Perplexity | 82 | $200,000 – $300,000 | $252,500 |
| 7 | Mercor | 90 | $130,000 – $300,000 | $238,750 |
| 8 | Replit | 58 | $200,000 – $277,500 | $236,250 |
| 9 | Waymo | 296 | $204,000 – $259,000 | $231,500 |
| 10 | Scale AI | 142 | $195,600 – $255,500 | $226,800 |
| 11 | Apptronik | 18 | $200,000 – $235,000 | $217,500 |
| 12 | Cognition | 12 | $180,000 – $237,500 | $201,250 |
| 13 | Agility Robotics | 65 | $170,000 – $235,000 | $198,000 |
| 14 | Figure AI | 61 | $150,000 – $250,000 | $197,500 |
| 15 | 1X | 66 | $174,430 – $232,500 | $195,000 |
Two things about the shape of this table. The gap between first and second is $55,000 — wider than the gap between second and seventh. And the widest single band belongs to xAI, whose postings run from $180,000 to $400,000 at the median, because that one board mixes data center technicians with frontier researchers.
The single highest range we found anywhere was Anthropic’s Engineering Manager, GPU at $500,000–$850,000. Eight Anthropic postings topped out at $850,000 on the day we read them. Its education line, verbatim: “Minimum education: Bachelor’s degree or an equivalent combination of education, training, and/or experience.”
Does it survive a like-for-like check?
A ranking of medians can be an illusion. If one company mostly posts principal researchers and another mostly posts field technicians, the first will “pay more” without paying any individual person more for the same work.
So we ran the ranking again on one narrow slice: postings whose title contains software engineer with no seniority word in it — no senior, staff, principal, lead, manager or director. Same job family, same rough level, nine companies with at least three such postings.
| Company | Non-senior SWE postings | Median midpoint |
|---|---|---|
| Anthropic | 9 | $362,500 |
| OpenAI | 129 | $317,500 |
| Mercor | 17 | $315,000 |
| xAI | 20 | $310,000 |
| Sierra | 22 | $310,000 |
| 1X | 7 | $240,000 |
| Replit | 6 | $215,000 |
| Scale AI | 12 | $202,500 |
| Waymo | 13 | $195,000 |
The order barely moves. Anthropic stays first at the same $362,500; OpenAI stays second. What does change is the middle: Mercor and Sierra climb, because their overall medians were being dragged down by non-engineering roles, and Waymo falls to last, because its large ops and operations-support population was propping the general number up.
So the headline holds. The top of this ranking is not a mix artifact.
The top two numbers are not the same kind of number
This is the part no salary listicle checks, and it took reading the fine print at the bottom of the postings themselves.
Four of the five biggest boards say plainly that the printed range is base salary and that equity sits on top of it. OpenAI’s Prototyping Lab Technician posting, verbatim: “In addition to the salary range listed above, total compensation also includes generous equity, performance-related bonus(es) for eligible employees, and the following benefits.” xAI’s technician posting: “Base salary is just one part of our total rewards package… which also includes equity.” Waymo prints “the expected base salary range” and lists its equity incentive plan separately. Scale AI: “Compensation packages at Scale for eligible roles include base salary, equity, and benefits.”
Anthropic’s wording is different. Its postings say “The annual compensation range for this role is listed below,” and the only time the word equity appears anywhere on the page is in the benefits line, as optional equity donation matching — a charitable-giving perk, not a grant.
What this does and does not mean
It does not mean the ranking is wrong. Anthropic posts the highest numbers, and that is a fact about the postings. It means the two numbers at the top of the table are labeled differently: four companies are showing you a floor with something else stacked on it, and one is showing you a figure it calls compensation. If you are comparing offers rather than postings, that distinction is worth more than the $55,000 gap. Ask which of the two a number is before you weigh it.
The best-paying company has no way in
Here is the finding that we think actually matters, and it runs the opposite direction to the ranking.
We counted, for each company, how many of its US postings are entry-level or hands-on technician work — the roles a person can reach without a research publication record. Then we put that next to the pay column.
Anthropic, first on pay, posted two technician roles out of 443 US openings and no entry-level ones at all — under half of one percent. OpenAI, second on pay, was at eleven out of 640. Sierra, fifth, had none of either.
Now look at the bottom of the pay column. 1X pays $195,000 at the median and 36% of its US postings are entry-level or technician work. Figure AI: 35%. xAI, which is fourth on pay, is at 26% — 61 openings out of 235.
The trade is roughly $170,000 of median posted pay against a door that is fifty times more likely to be open. That is not a small difference in strategy; it is a different question with a different answer. If you came here to benchmark an offer, read the left column. If you came here because you want one of these jobs, read the right.
A quarter of these jobs never say
Of the 2,936 US postings we read, 2,144 printed a range and 792 did not. That is 73% transparency overall, and it is distributed very unevenly.
ElevenLabs, Cursor and Physical Intelligence published a range on zero of their 231 US postings between them. Nuro managed three out of 101. Those four are absent from our ranking not because they pay badly — we have no idea what they pay — but because they did not say.
The reason is mostly law rather than temperament. California, Washington, New York, Colorado and a growing list of other states require a pay scale in the posting itself; most states do not. We took that apart on a single company’s board in what xAI actually pays in Memphis, where the same employer published a range on 96% of its California postings and 3% of its Tennessee ones.
Which is the quiet reason to be skeptical of every “highest paying AI companies” list you find today, including this one: any such ranking can only describe the companies that answered.
What the reachable jobs actually pay
The six-figure medians above belong overwhelmingly to senior engineering and research roles. The natural follow-up question — what do the jobs I could actually get pay? — has a much thinner answer, and the thinness is itself the finding.
We found 77 entry-level US postings across all twenty boards. Only ten of them published a number. Here are all ten, every link going to the employer’s own page.
| Role | Employer | Location | Published range |
|---|---|---|---|
| Research Fellowship — APEX | Mercor | Remote (SF) | $40,000 – $80,000 |
| Senior Service Technician II | Agility Robotics | Salem, OR | $60,000 – $78,000 |
| IT Services Technician | xAI | Memphis, TN | $61,200 – $91,800 |
| IT Services Technician | xAI | Seattle, WA | $68,400 – $102,600 |
| Engineering Support Technician | Agility Robotics | Fremont, CA | $74,000 – $95,000 |
| Senior Technician, Robotics | Scale AI | San Francisco, CA | $134,400 – $168,000 |
| Datacenter Networking Technician | OpenAI | San Francisco, CA | $157,000 – $221,000 |
| Site Operations Technician | OpenAI | San Francisco, CA | $157,000 – $221,000 |
| ML Engineer, Perception (New Grad) | Waymo | Mountain View, CA | $170,000 – $216,000 |
| Prototyping Lab Technician, Robotics | OpenAI | San Francisco, CA | $225,000 – $295,000 |
The median midpoint of those ten is $118,350. But notice how little that median means: the list runs from a $40,000 remote fellowship to a $225,000–$295,000 lab technician job at the same seniority label, and seven of the ten are technician work rather than desk work. The label “entry-level” is doing almost no work here. The specific posting is doing all of it.
We walked the route into the largest of these categories in how to become a data center technician with no experience, and the robotics equivalent in do you need a degree to be a robotics technician.
What moved this week
We run this scan every day, so we can say what changed rather than guess. Between Saturday, August 22 and Saturday, August 29, the twenty boards went from 3,645 open roles to 3,716. That net gain of 71 hides a lot of churn: 322 postings appeared and 251 came down.
| Company | Posted | Closed | Net |
|---|---|---|---|
| Anthropic | +98 | −39 | +59 |
| Scale AI | +14 | −3 | +11 |
| Cursor | +9 | −1 | +8 |
| Sierra | +9 | −3 | +6 |
| Apptronik | +8 | −3 | +5 |
| Cognition | +6 | −1 | +5 |
| OpenAI | +69 | −65 | +4 |
| Nuro | +10 | −13 | −3 |
| Cohere | +7 | −11 | −4 |
| xAI | +16 | −21 | −5 |
| ElevenLabs | +17 | −23 | −6 |
| Waymo | +19 | −33 | −14 |
The week’s biggest hiring move belongs to the same company that tops the pay table: Anthropic added 98 postings and removed 39. Waymo went the other way, closing 33 and opening 19. OpenAI posted the most raw volume of anyone — 69 new roles — and still finished the week up only four, which tells you how fast that board recycles.
The running totals update daily on the AI Jobs Tracker.
What to do with this
- Decide which column you are shopping in first. Pay and access are inversely correlated on this list. Chasing the top of the pay table with a non-research background is the most common way to spend six months getting nowhere.
- Before you compare two numbers, find out what each one includes. Scroll to the bottom of the posting. If it says “in addition to the salary range… equity,” you are looking at a floor. If it says “annual compensation range,” you may not be.
- If the posting has no number, that is usually geography, not stinginess. Look up the same title on the same company’s California, Washington, New York or Colorado listing. Same employer, same band, printed because the law says so.
- Apply to the specific posting, not the company. Seven of the ten published entry-level ranges above are technician roles, and they are on the same boards as the $850,000 engineering manager job. The board is not one market.
- Check before you apply. Every figure here was read on August 29, 2026. These postings close without notice and the employer’s own page is the only thing that counts.
If you want to go deeper
“Which of these companies is actually hiring the most people?” Not the same question as who pays most — OpenAI, Anthropic and Waymo carried the largest US boards on the day we counted. The breakdown is in which AI companies are hiring right now.
“How do I get hired at the top of that table?” It is a narrow and specific path, and we wrote it out in how to get a job at OpenAI or Anthropic.
“Are there employers who publish everything?” Yes — federal agencies and the national laboratories, by rule. See how to get a federal job with no experience and national laboratory jobs, no degree.
“What about the remote AI work I keep seeing advertised?” Some of it is real and some of it is not. We checked in are AI training jobs legit. For the anxious version of the whole subject, will AI take my job, and for the physical side of the industry, our Physical AI library.
Have a question about a specific company, city or role? Ask the desk — we answer on the AI Jobs Desk page, and we check the employer’s own page before we reply. If your question is broader than jobs, Ask Prof H.
Questions people actually search
Which AI company pays the most?
On August 29, 2026, Anthropic published the highest salary ranges of any AI or robotics company we track: a median posted US range of $315,000–$405,000, midpoint $362,500, across 427 postings. OpenAI was second at $250,000–$345,500 (midpoint $307,500) across 612 postings, then Cohere at $280,000 and xAI at $275,000. The ranking holds when you narrow it to non-senior software engineering roles only, so it is not an artifact of who posts more senior jobs.
What is the highest paying AI job posting right now?
The highest range we found on August 29, 2026 was Anthropic’s Engineering Manager, GPU (ML Accelerator) in San Francisco at $500,000–$850,000. Eight separate Anthropic postings topped out at $850,000 that day, including Performance Engineer, Pre-training Distributed Systems Tech Lead, and Research Engineer for Chip Design RL. Its stated minimum education is a bachelor’s degree or an equivalent combination of education, training and experience.
Do these salary ranges include equity?
It depends on the company, and the wording matters. OpenAI, xAI, Waymo and Scale AI all state in the posting that the range is base salary and that equity is granted on top of it. OpenAI’s wording is “In addition to the salary range listed above, total compensation also includes generous equity.” Anthropic describes its figure as “the annual compensation range for this role” and does not say equity is additional. So the two numbers at the top of the ranking may not be measuring the same thing, and you should ask which kind of number you are being shown before comparing offers.
Which AI company is easiest to get hired at without a degree or experience?
By share of openings rather than reputation, the humanoid robotics companies. On August 29, 2026, 36% of 1X’s US postings and 35% of Figure AI’s were entry-level or hands-on technician roles, and xAI was at 26% (61 of 235). At the other end, Anthropic listed two technician roles out of 443 US postings and no entry-level ones, and Sierra listed none of either. Pay and access ran in opposite directions across the whole list.
How much do entry-level AI jobs pay?
There were 77 entry-level US postings across the twenty boards on August 29, 2026, and only ten published a figure. Those ten ran from $40,000–$80,000 (a remote Mercor research fellowship) to $225,000–$295,000 (an OpenAI robotics prototyping lab technician in San Francisco), with a median midpoint of $118,350. Seven of the ten were technician roles rather than desk jobs, so the specific posting matters far more than the entry-level label.
Why do so many AI job postings not show a salary?
Mostly because in most states they are not required to. California, Washington, New York and Colorado, among others, require a pay scale in the posting itself; most states do not. Across the 2,936 US postings we read on August 29, 2026, 2,144 printed a range and 792 did not — 73% transparency. ElevenLabs, Cursor and Physical Intelligence published a range on zero of their 231 US postings between them, and Nuro on three of 101. Any ranking of AI pay, including ours, can only describe the companies that answered.
Cite this
profhlab.com, “Which AI Company Pays the Most? We Read 2,144 Posted Salary Ranges,” August 29, 2026. Anthropic led with a median posted US range of $315,000–$405,000 (midpoint $362,500) across 427 postings; 73% of 2,936 US postings on twenty AI and robotics boards published a range. Method: How We Count. Daily totals: AI Jobs Tracker.
Sources: the public job boards of twenty AI and robotics employers, read August 29, 2026; individual postings linked above. Images: Marinus van Reymerswaele, A Moneychanger and His Wife (public domain); Oak Ridge National Laboratory (CC BY 2.0); U.S. Army Corps of Engineers Sacramento District (public domain); 9yz (CC BY 4.0), via Wikimedia Commons. New to the site? Start here, or browse the FAQ. One short email a month.
