U.S. Tech Giants Rethink AI as Token Costs Surpass Programmer Wages

The AI industry today presents a stark contrast, with extremes on both sides. Major AI companies have surpassed the trillion-dollar valuation mark, and large language models are now widely used in coding. Yet many businesses that hoped to cut labor costs through AI have been surprised to discover that it actually ends up being more expensive than hiring human programmers.
This unexpected turn has left numerous tech companies in an awkward position. AI coding tools, once expected to lower costs and boost efficiency, now generate enormous bills for finance departments due to their incredible consumption speed.
Even Major Companies Must Reckon with Skyrocketing Budgets
Uber's chief technology officer recently acknowledged that the company had already exhausted its entire 2026 Claude Code budget by April. To cover these excessive AI costs, Uber had to slow down its annual hiring plans—prompting management to deeply rethink the blind rush toward internal AI adoption.
Unsurprisingly, cash-rich Microsoft is also grappling with the same financial drain. CEO Satya Nadella recently ordered a switch from Claude Code to the company's own GitHub Copilot for internal development starting in June. The core intention: strictly control spending and prevent runaway costs from large models.
24/7 Operation Turns into a "Money-Guzzling Beast"
While AI programming tools write faster than humans, in practice human developers earning a few thousand dollars per month are often more cost-effective than token-based AI. Particularly when teams deploy AI agents to run around the clock, money drains away like water released from a dam.
Beyond the heavy financial burden, the quality of AI-generated code also faces industry criticism. Several experts note that AI often "mass-produces garbage," with many hidden defects in the output. Subsequent review, testing, and deployment still require human cleanup. Replacing human programmers in the near term remains an unfulfilled promise.
Related article
Midjourney Launches Full-Body Ultrasound Scanner, Expanding Beyond Generated Images
Leading AI image platform Midjourney has expanded into healthcare, launching its first hardware device: a full-body ultrasound scanner. Named the Midjourney Scanner, it uses an ultrasound ring sensor to capture 3D internal images with quality rivalin
WordPress Parent Company Launches Android Version of Mesh, Expanding AI Assistant to Cross-Platform Nexus
Mesh, Automattic’s personal CRM and relationship management platform (formerly Clay), officially launched on Android on August 12, 2026. The app supports multi-screen adaptation, pop-up views, foldable devices, and real-time cross-device synchronizat
ChatGPT User Base Nears 1 Billion After Seven-Month Delay Amid Emerging Challenges
Internal data from OpenAI reveals that ChatGPT’s weekly active users are nearing the 1 billion mark. This milestone, initially projected for late last year, was achieved seven months later.Despite the delay, this scale remains impressive. In under fo
Related Special Topic Recommendations
Comments (0)
0/500

The AI industry today presents a stark contrast, with extremes on both sides. Major AI companies have surpassed the trillion-dollar valuation mark, and large language models are now widely used in coding. Yet many businesses that hoped to cut labor costs through AI have been surprised to discover that it actually ends up being more expensive than hiring human programmers.
This unexpected turn has left numerous tech companies in an awkward position. AI coding tools, once expected to lower costs and boost efficiency, now generate enormous bills for finance departments due to their incredible consumption speed.
Even Major Companies Must Reckon with Skyrocketing Budgets
Uber's chief technology officer recently acknowledged that the company had already exhausted its entire 2026 Claude Code budget by April. To cover these excessive AI costs, Uber had to slow down its annual hiring plans—prompting management to deeply rethink the blind rush toward internal AI adoption.
Unsurprisingly, cash-rich Microsoft is also grappling with the same financial drain. CEO Satya Nadella recently ordered a switch from Claude Code to the company's own GitHub Copilot for internal development starting in June. The core intention: strictly control spending and prevent runaway costs from large models.
24/7 Operation Turns into a "Money-Guzzling Beast"
While AI programming tools write faster than humans, in practice human developers earning a few thousand dollars per month are often more cost-effective than token-based AI. Particularly when teams deploy AI agents to run around the clock, money drains away like water released from a dam.
Beyond the heavy financial burden, the quality of AI-generated code also faces industry criticism. Several experts note that AI often "mass-produces garbage," with many hidden defects in the output. Subsequent review, testing, and deployment still require human cleanup. Replacing human programmers in the near term remains an unfulfilled promise.
Midjourney Launches Full-Body Ultrasound Scanner, Expanding Beyond Generated Images
Leading AI image platform Midjourney has expanded into healthcare, launching its first hardware device: a full-body ultrasound scanner. Named the Midjourney Scanner, it uses an ultrasound ring sensor to capture 3D internal images with quality rivalin
WordPress Parent Company Launches Android Version of Mesh, Expanding AI Assistant to Cross-Platform Nexus
Mesh, Automattic’s personal CRM and relationship management platform (formerly Clay), officially launched on Android on August 12, 2026. The app supports multi-screen adaptation, pop-up views, foldable devices, and real-time cross-device synchronizat
ChatGPT User Base Nears 1 Billion After Seven-Month Delay Amid Emerging Challenges
Internal data from OpenAI reveals that ChatGPT’s weekly active users are nearing the 1 billion mark. This milestone, initially projected for late last year, was achieved seven months later.Despite the delay, this scale remains impressive. In under fo





Home






