What AI development costs in Silicon Valley

A buyer will want to know the price. For the fourteen firms in this guide, there is no public answer. None of them publishes a price on the pages read for this guide on 20 October 2026. Every quote is written for one project.

That leaves two things a buyer can use: public wage data, which shows what the people who build software are paid here, and a careful way to compare the quotes you get.

No published prices

Firm	Published price	What to ask for
Reveneau	None	A written quote; Reveneau says it puts the saving from using AI in place of more engineers into the price
Fluxon	None	A written quote with a breakdown by phase
Ode with Anthropic	None	A written quote with a breakdown by phase
Distyl AI	None	The cost of the work and of any Distyl product used
Tribe AI	None	Rates for employees and for contractors
Loka	None	The cost of the work and of the AWS services used
Provectus	None	The cost of the work and of any packaged system used
Factored	None	Rates per engineer, and the notice period
ClearScale	None	The cost of the work and of the AWS services used
Thoughtworks	None	Rates by role, and the cost of any Thoughtworks platform
Slalom	None	Rates by role, and the cost of running the system after launch
Grid Dynamics	None	Rates by role and by location
Turing	None	The cost of the work and of the Turing platform
IDEO	None	The cost of the design phase, and who builds afterwards

Several firms sell a product or platform alongside the work, by their own accounts: Distyl deploys its own products, Thoughtworks sells AI/works and Agent/works, Turing offers an agent management platform, and Provectus describes packaged systems for its two industries. Ask for those costs as separate lines, because they may continue after the project ends.

What engineers are paid here

The US Bureau of Labor Statistics publishes wage estimates for every occupation and metro area. Its May 2025 figures for software developers, read through the BLS data service, are below [1] [2] [3] [4] [5] [6] [7] [8]. The occupation is the one BLS calls software developers. This guide uses it because it is the closest occupation to the people who build AI software.

Area	Median annual wage	Mean annual wage	Software developers employed
San Francisco, Oakland, Fremont metro area	$186,640	$193,430	69,030
San Jose, Sunnyvale, Santa Clara metro area	$213,110	$221,710	87,350
United States	$135,980	$148,100	Not used here

The median is the middle wage: half of developers earn more, and half earn less. The mean is the average. In all three rows the mean is higher than the median. That happens when the highest wages sit further above the middle wage than the lowest wages sit below it.

The differences are simple to compute. The San Francisco median is $50,660 above the national median. The San Jose median is $77,130 above it, and $26,470 above San Francisco's.

The same wages per hour

A full-time year is often counted as 2,080 hours: 40 hours a week for 52 weeks. Dividing each annual figure by 2,080 gives an hourly wage. This is this guide's own division of the BLS figures.

Area	Median per hour	Mean per hour
San Francisco metro area	$89.73	$93.00
San Jose metro area	$102.46	$106.59
United States	$65.38	$71.20

These are wages: what an employer pays the developer. A firm's charge to a client is a separate number, covered next.

What a firm's price covers beyond wages

A firm that sells engineering time pays its engineers' wages, and it also pays for everything else it needs to operate: benefits, offices, managers, sales, tools, and the weeks when an engineer is between projects. Then it adds a profit. So a rate charged to a client is higher than the wage of the engineer doing the work.

How much higher depends on the firm, and none of the firms in this guide publishes the figure. Ask each firm to break its quote into the roles on the project and the rate for each. Then you can see how the rate relates to the wage data above.

Wages for a team, as an example

To show how wages add up, the table below multiplies the median annual wage by the size of a team. It shows wage cost only, as an example, and it leaves out everything in the section above. No firm quoted these figures.

Team size	San Francisco median wages	San Jose median wages	US median wages
2 developers	$373,280	$426,220	$271,960
4 developers	$746,560	$852,440	$543,920
6 developers	$1,119,840	$1,278,660	$815,880

The point of the table is the size of each step. Each developer added to a San Francisco team adds $186,640 a year in median wages alone.

Where AI changes the cost

That step is why the number of people on a project affects its price. Reveneau says it uses AI in place of hiring more engineers, and it puts that saving into the price clients pay. By its own account, AI writes 100 percent of its code, and every change has to pass a large eval suite written from the client's specification before release, graded by a model called Jev. Reveneau says Jev makes that suite run ten times faster than its previous grader did. Reveneau takes responsibility for the whole project through production and after release. To see what this means for your project, ask Reveneau for a quote. Its AI development service describes the work.

Ask every other firm the same question: how does your use of AI show up in my quote? A firm should be able to answer in specific terms.

Costs that continue after launch

An AI system has costs after it is built. Ask about each of these before you sign.

• Model use. A system that calls a model maker's model pays the model maker for that use. Ask the firm to estimate this from your expected use, and to show the working.
• Hosting. The servers, databases and storage the system runs on.
• Platform fees. Any vendor product the system depends on, charged monthly or yearly.
• Changes. Models change, data changes, and the business changes. Ask what a month of maintenance costs, and what it includes.

How to compare quotes

The US Office of Management and Budget tells federal agencies to state in their requests that they want AI proposals that reduce the risk of vendor lock-in, which means being unable to leave a vendor without losing your work. Clear pricing is one of the four protections it names, with knowledge transfer, clear rules for moving data and models, and clear licence terms [9]. A company can ask for the same.

1. Send the same brief to every firm. If the briefs differ, the quotes cannot be compared.
2. Ask for roles, rates and hours. A single total hides how the firm plans the work.
3. Ask for running costs separately. Model use, hosting and platform fees.
4. Ask what is not included. Ask in particular whether data cleaning, security review and training for your staff are in the price.
5. Ask what changes the number. A good quote names the events that would raise it.
6. Pay for a small test first. A short paid task on your own data shows how a firm works before a large commitment.

The questions to ask an AI development company page includes these in a printable checklist, and the full list has each firm's entry.

What the wage figures leave out

The BLS figures describe one occupation, software developers, in May 2025, for each metro area as BLS defines it. A wage is the pay an employee receives, and an employer's costs for benefits, offices and other items come on top of it. So use the figures as a starting point for questions, and ask each firm for its own rates.

One practical use: divide a firm's hourly rate for a role by the hourly median wage for the area where that person works. The result shows how many times the median wage the firm charges for that role. Ask the firm to explain the difference, role by role.

Where the engineers work

Wages differ by place, as the tables show. The San Jose metro median is $77,130 above the national median, and the San Francisco metro median is $50,660 above it. Some firms on this list name places outside the Bay Area where they work. Provectus lists engineering hubs in eight countries, and Thoughtworks lists 47 offices in 18 countries, by their own accounts. Ask each firm where the engineers on your project will work, and whether its rates change with the location.