Nika

How will big companies deal with the cost of AI?

A lot of large companies (not just tech giants) are laying off employees at scale because AI can already do parts of their work faster, in higher volume, and at a comparable level of quality, sometimes even better.

At the same time, companies are starting to openly admit something important: AI is expensive.

Not just in terms of infrastructure, but also energy consumption and operational costs.

For example:

→ Uber reportedly deployed AI across its 5,000 engineers, and within 4 months, they exhausted their entire annual budget. Their COO admitted they can’t justify the expense.

→ Microsoft has revoked AI licenses from its own developers to control costs.

→ Starbucks scrapped its AI inventory system after 9 months because it performed worse than human employees.

→ NVIDIA’s VP recently said that “AI is costing more than human workers.”

So the question is (are):

  1. Will companies eventually find a way to significantly reduce the cost of AI?

  2. Or will we see a partial return to “old-school” human labour because it’s still more reliable and cost-efficient in certain cases?

  3. What’s your take on this?

84 views

Add a comment

Replies

Best
Jiaqi Lu
I work at a global tech enterprise, and we are seeing a mixed approach to this issue. In our company, AI tooling budgets are managed at the team level. For example, our team has a fixed monthly cap of $50 per developer for Cursor, while some other teams I know have unlimited credits for AI tools. Because of this, we naturally adopted a hybrid workflow: for quick, straightforward tasks, we go old-school to stretch our budget. We save the precious credits for complex bugs or stuff that we are not familiar with. It really is a balancing act between efficiency and cost right now.
Nika

@jiaqi_lu Do you have any tracking system that estimates what is cheaper? (human vs. AI)

Jiaqi Lu
@busmark_w_nika I’m not really sure coz I’m just a engineer, I’m not in the manager level. So I’m not sure if they are tracking it. But I think our company is still not that “aggressive” about the AI tools. It’s still a tool for helping us develop, but we still need to take our time to understand the code.
Adela C

I think it will eventually settle into a middle ground, where AI is used as a tool, enabling small(er) teams of humans to perform better. I'm not sure the cost will be driven down too soon, given how resource hungry AI infrastructure is proving to be

Nika

@adela_c Time will show us :)

Fatima Khan

There is no denying the use of AI, however its a steep hill if not managed properly.
I think once the hype dies down, companies will figure out how to reduce the cost and go old school as well.
A balance that will often tip either way.

Nika

@fatima_shehzad I sometimes have in my mind that Terminator movie that the only way to fight AI and robots is in the war against them (humans vs machines)

Fatima Khan

@busmark_w_nika That thougth has often crossed my mind. What if we were pre warned about AI through these movies. I, Robot, Robo Cop, A.I and so many more point towards something.

Nika

@fatima_shehzad yeah, those 90s movies too :D

Farrukh Butt

I don’t think companies will fully go back, but I do think the “AI everywhere” phase will cool down. Once the bill shows up, the companies will start asking where AI actually saves money and where it just adds another expensive layer. Some work is still cheaper and cleaner with humans.

Nika

@farrukh_butt1 But it still feels like people are burning money on those AI startups and VCs are still okay to fund them.

İlknur B.

I don't think it will be a simple AI vs human decision. Companies will probably use AI more selectively, smaller models, limits, routing and humans for the parts where judgment still matters. The hard part is measuring real ROI, not just usage. A tool can look productive and still be expensive in the wrong workflow.

Nika

@bailknur Okay, so in which areas or when AI will NOT be used? :)

Zhen Han
Use codex/claude for coding, open source models for everything else.
Nika

@zhen_han What solution are you coding with it? Volak or any other too?

Zhen Han

@busmark_w_nika Vokal is pretty new. I first built my own multi-agent setup in the terminal, and my teammate built theirs separately. We eventually converged on the same pattern and started building Vokal about 3 months ago. Since then, our team has been doing most of our work on Vokal.

It does not have to be Vokal if you are doing single-player multi-agent coding and staying very hands-on. But for multi-agent teams, Vokal is definitely a productivity boost because it keeps the collaboration, handoffs, approvals, and history traceable in one place.