offshoring on S Anand

Sat, 24 May 2025 07:58:38 +0000

How much does an LLM charge per hour for its services?

If we multiple the Cost Per Output Token with Tokens Per Second, we can get the cost for what an LLM produces in Dollars Per Hour. (We’re ignoring the input cost, but it’s not the main driver of time.)

Over time, different models have been released at different billing rates.

New powerful models like O3 cost ~$7/hr – Poland’s minimum wage rate.
Gemini 2.5 Pro costs ~$12/hr – France’s minimum wage rate.
The latest Claude 4 Sonnet costs ~$2/hr – India’s minimum wage rate.

Once a language model’s run-time cost drops below the local minimum wage, the “offshoring” advantage disappears. AI becomes the cheapest employee in every country at once.

Paradoxically, workers in countries with strong labor protections, unions, and higher wages (like Germany and France) may be safer from AI displacement.

But countries whose economies depend on wage arbitrage face economic disruption.

Visualization: https://public.flourish.studio/visualisation/23384855/
Analysis: https://chatgpt.com/share/68317a06-0cac-800c-ad6f-13646ceb489f
Source code: https://github.com/sanand0/llmpricing/blob/9b66480458b712de52ea14502a64bf4c3b6a98ed/billing.py

Wage Rates of Nations and LLMs

Sat, 24 May 2025 07:47:01 +0000

How much does an LLM charge per hour for its services?

If we multiple the Cost Per Output Token with Tokens Per Second, we can get the cost for what an LLM produces in Dollars Per Hour. (We're ignoring the input cost, but it's not the main driver of time.)

Over time, different models have been released at different billing rates. Most new powerful models like O3 or Gemini 2.5 Pro cost ~$7 - $11 per hr.

To get a sense of this, let's look at wage rates across countries and industries:

Rate ($/hr)	Countries (avg hourly wage)	Models
0–2	Bangladesh ($1.42/hr), Pakistan ($1.65/hr), Vietnam ($0.94/hr)	devstral-small ($0.01/hr), gemini-2.5-flash-preview ($1.50/hr)
2–5	Brazil ($3.09/hr), Mexico manufacturing ($4.90/hr)	claude-sonnet-4 ($2.23/hr), codex-mini ($2.54/hr)
5–10	India ($5.03/hr), South Africa avg ($9.38/hr), Poland min wage ($7.35/hr)	o3 ($7.16/hr), claude-opus-4 ($8.67/hr)
10–15	Germany ($12.93/hr), France ($12.41/hr), UK ($14.43/hr)	gemini-2.5-pro-preview ($11.89/hr), gpt-4.5-preview ($13.10/hr)
15–20	Spain ($15.87/hr), Italy ($16.80/hr), Japan ($17.98/hr)	No recent models in this range

Workers in Europe and Japan are already more expensive than the more expensive models, at $12+ per hour. India, Brazil, Mexico etc. are more expensive than most of the average models.

Once a language model’s run-time cost drops below the local minimum wage, the “offshoring” advantage disappears. AI becomes the cheapest employee in every country at once. Countries whose economies depend on being the "cheaper alternative" for labor-intensive work face potential economic disruption.

Paradoxically, workers in countries with strong labor protections, unions, and higher wages (like Germany and France) may paradoxically be safer from AI displacement.

Source code: sanand0/llmpricing
Analysis: ChatGPT

Comments

Soumendra Kumar Sahoo 24 May 2025 1:44 pm: The white collar workers jobs are in danger compared to blue collar ones.

Change management

Sun, 30 Apr 2006 12:00:00 +0000

Change management can be analytic, as opposed to touchy-feely.

Our client’s operating margin was falling. The bosses wanted to offshore their back office. Others weren’t convinced. To manage this change, we needed three questions answered:

Who’s not convinced?
Why aren’t they convinced?
What’ll convince them?

Who’s not convinced?

We plotted the level of support and importance of key people on the stakeholder support matrix. This split people into 4 groups (below). Then we showed it around to people and had them move people around on the matrix.

Minor sceptics. We largely ignored them
Change facilitators. We tasked them with roles in the project
Change agents. We made them influence the others
People to convince. The ones we needed to focus on

This is a simple concept, actually. The insight is, putting names on such a matrix, and getting people to move them around, is a robust way to get everyone on the board and at the right spot.

Why aren’t they convinced?

We sent everyone a list of benefits and issues in outsourcing. They rated them. We grouped the results and plotted them. Here’s the result for Uli.

Uli saw more issues than benefits. Quality and possible better service were benefits. But he was afraid the company wasn’t ready, and vendors wouldn’t understand their operations.

The advantage of these charts is that you can put them side by side, and compare where different people stand. It gives you a great view of why they’re objecting, and whom you can use to counter that.

What’ll convince them?

Once we knew why people objected, it was easy to manage.

For example, to counter Uli’s fear of organisational readiness, we got people who felt this was not an issue to put forward their counterpoints.

To counter fears of vendor ability, we got a bunch of them to visit BPOs in India, and spread their confidence to others.

We arranged workshops, making sure that each group had people to convince and change agents.

This did require a lot of soft skills. But the success was largely because of the structured ground-work. Change management can be quite analytic.

Comments

ravi 1 May 2006 4:12 pm: cool work..something similar to stakeholder mapping rite.but how did u achieve the moving around of people within the matrix .was it a group exercise.is there any chance of a bias entering the analysis ? how did u arrive at the final stakeholder map? i did a study for scorttish power on wind power..we didnt exactly conclude the arguement as i didnot know how to achieve the support of the stkeholders or the OR way of achieving it..this could have surely helped if it had come 10 days ago anand..really cool
Sai 1 May 2006 11:36 pm: Excellent. Soft-skills help a lot, but this kind of effort goes a long way in assuaging skeptics. Thanks for sharing.
Arun 2 May 2006 6:16 am: Very neat. I am going to try and use this if the opportunity presents itself. Thanks! Btw, can you share a little more info on the company that’s using the wiki?
S Anand 3 May 2006 7:13 am: Arun, I’m afraid I don’t know much more about them myself. Just that there are examples that my colleague vouches for. Infosys is going for an internal wiki initiative as well.
Sumeet 3 May 2006 2:51 pm: The octagonal chart is confusing at first. Isn’t there a better way of plotting the issues/benefits poll results ?
S Anand 3 May 2006 5:58 pm: Definitely! Variants of the bar chart could have done the job much better. We were pressed for time, though, and we felt cool showing spider graphs to the client… :-)
Arun 4 May 2006 5:29 am: Ah, alright. Wipro had tried blogs for a while, i think it died out pretty soon. Interesting, infy going for a wiki initiative. Hope you guys fare better! :-)
Jayant 4 May 2006 1:41 pm: Honestly, I dont have numbers to prove but I feel this outsourcing is breaking into other’s “rice bowl”…but change is change..
Sumeet 4 May 2006 3:21 pm: What strategy did u use for the people on the middle ground. In your e.g Jim K and Mark L can go either way (People to convince or change agents) ?
S Anand 4 May 2006 8:21 pm: Jayant, you’re right. When we were working on this case, we went out to lunch at a restaurant. Behind us were 5 people who’d been laid off by this client 6 months ago. I overheard their conversation. They still hadn’t found a job. Even daily living was proving tough. There’s no doubt: outsourcing hits a lot of people directly, and hard.
S Anand 4 May 2006 8:21 pm: Sumeet, Jim H and Jim K were the top people in the company. We didn’t want to “counter” them with anyone having strong opinions. It was vital not to THRUST anything on these two. We asked them for their initial reactions and questions, went back, spent a LOT of time preparing answers, and had conversations with them. We moved gently, and gave them a lot of time. And that’s where the soft stuff comes in.
jayant 6 May 2006 2:41 pm: Actually there is interesting parallel to what happened to Indian textile industry in early 1900. Powerloom’s shirts were cheaper than handloom’s. So, Indian weavers were losing jobs to English powerloom workers. Not that powerloom workers were well off. You may want to read Autobiography of Gandhi for that. Interestingly, then there was Swadeshi movement etc and Indian business community infact had said that it does not make sense to stop imports as it cheaper and people will buy what is cheap and good. Lot of parallels actually. Anyway, guess people who moved to West and probably toiled are probably better off. Hell for one generation, probably heaven for next!
S Anand 6 May 2006 3:42 pm: Hey, that’s an interesting perspective!
damand draft 11 Dec 2006 3:57 am: deamand draft