OpenAI's infrastructure spending projections have become a focal point in the AI industry, with Bernstein analyst Mark Moerdler estimating commitments could reach $614 billion through 2030. This figure falls below the widely cited $750 billion estimate, as delays and cancellations may reduce the final amount. The analyst projects annual spending could peak at $184 billion in 2031, while OpenAI itself reached nearly $50 billion in annualized revenue by September, representing 77 percent year-over-year growth. These figures highlight the scale of investment required to support advancing AI capabilities while the company works toward sustainable revenue generation.
Capital Economics has raised concerns about the sustainability of the current AI investment boom, noting that the five largest US hyperscalers plan to spend $800 billion this year on capital expenditure, with capital expenditure, a figure that could rise to $1 trillion or more by 2027. However, the firm warns AI-related revenues would need to grow nine to twelvefold by 2030 to justify such spending levels. The analysis predicts investment growth will slow in 2027 and decline approximately 10 percent in 2028, with potential consequences including reduced US GDP growth and impacts on exports in Taiwan and South Korea. These warnings come amid market scrutiny of whether current spending levels can be sustained given revenue timelines.
Market dynamics continue to shift across related sectors. Winbond Electronics chairman Chiao Yu-Chun indicated memory prices may have peaked, citing the company's October revenue of NT$209.79 billion for January through October, up 195.45 percent from the same period in 2025. Chiao believes memory makers have made considerable profits and prices should not rise further, with the next market downturn expected after 2029 when funding limits potentially tip the market. Meanwhile, US export controls introduced in October led to a 20 percent decline in global semiconductor sales to China according to IC Insights, with most of the decline occurring in the third quarter and sales also falling to South Korea and Taiwan.
On the productivity front, a study of over 9,100 employees at more than 450 companies from April to September revealed heavy AI users work longer days, averaging between 8.4 and 9.1 hours compared to 7.6 hours for non-AI users. Researcher Gabriela Mauch noted AI may help complete tasks faster but contributes to longer workdays and more application switching, with heavy AI users experiencing more interrupted work than light or non-AI users. The findings suggest AI's impact on workplace productivity presents complex trade-offs as organizations adapt to new working patterns.
Upwork reported third-quarter revenue of $153.8 million, up 22 percent from the same period last year, posting net income of $5.4 million or $0.06 per share compared to a net loss of $1.4 million in the comparable period. CEO Hayden Barnes cited strong demand for AI products and enterprise services as the company's stock price rose 12 percent to close at $13.49 per share with a market capitalization over $1 billion. These results demonstrate continued business adaptation to AI integration while maintaining financial performance.
Kepler CEO Ganesh emphasized that lookup must perform better than search to build trust in AI systems, describing a two-stage process that pulls up to 200 candidates with ANN and BM25, then narrows to 60 using ranking and cross-encoders. The system blocks access from other firms using tenant scoping and allow lists, with Ganesh citing data showing 80 percent of finance executives trust numbers while 31 percent find errors. This focus on reliability and trust-building represents an important dimension of AI deployment beyond pure capability improvements.
Hancom faces governance scrutiny ahead of its AI service launch, with the company planning to release its AI service beta in the U.S. market next month and launch globally in March. Chairman Kim has served as an in-house director since 2010 and was fined 20 million won for failing to report changes in his stake in Hancom With. Experts say Hancom needs to reform its board and increase independent directors to become a trusted global company, highlighting the importance of corporate governance in the AI sector's expansion into new markets.
eGain plans to increase sales and marketing spending in fiscal 2027, targeting about 20 percent growth in AI customer ARR by June 30, 2027. The company reiterated its fiscal 2027 AI customer revenue outlook of $59.5 million to $60.5 million, though it cautioned that first-quarter profit performance does not represent expectations for the rest of the year. This forward-looking approach reflects ongoing investment in AI customer solutions despite near-term profit variability.
The FT's Lex column warns that AI will reduce the number of managers companies need, potentially lowering pay for managers who are often the highest-paid employees. Companies will need to adapt their structure and become more efficient as this shift may reduce overall job numbers and impact the broader economy. This prediction adds to growing discussions about AI's societal implications beyond productivity and market dynamics.
Key Takeaways
- Bernstein analyst Mark Moerdler projects OpenAI's AI computing commitments could reach $614 billion through 2030, lower than the widely cited $750 billion estimate.
- OpenAI reached nearly $50 billion in annualized revenue by September, representing 77 percent year-over-year growth.
- Capital Economics forecasts the five largest US hyperscalers will spend $800 billion this year on capital expenditure, potentially rising to $1 trillion or more by 2027.
- AI-related revenues would need to grow nine to twelvefold by 2030 to justify current spending levels according to Capital Economics analysis.
- US export controls introduced in October led to a 20 percent decline in global semiconductor sales to China, with most decline occurring in the third quarter.
- Heavy AI users work between 8.4 and 9.1 hours daily compared to 7.6 hours for non-AI users, according to a study of over 9,100 employees.
- Winbond Electronics reported October revenue of NT$209.79 billion, up 195.45 percent from the same period in 2025, with chairman Chiao hoping memory prices have peaked.
- Upwork posted Q3 revenue of $153.8 million, up 22 percent year-over-year, with net income of $5.4 million or $0.06 per share.
- Kepler uses a two-stage lookup process that retrieves up to 200 candidates with ANN and BM25, then narrows to 60 using ranking and cross-encoders to build AI trust.
- Hancom faces governance scrutiny with Chairman Kim fined 20 million won for failing to report stake changes as the company prepares its AI service beta launch.
OpenAI compute spending lower than expected, analyst says
OpenAI's AI computing commitments could lead to $614 billion in infrastructure spending through 2030, according to Bernstein analyst Mark Moerdler. That figure is lower than the widely cited $750 billion estimate. Delays and cancellations could reduce spending further. Annual spending may peak at $184 billion in 2031. OpenAI reached nearly $50 billion in annualized revenue in September, growing 77 percent year over year.
OpenAI compute spending lower than expected, analyst says
OpenAI's AI computing commitments could lead to $614 billion in infrastructure spending through 2030, according to Bernstein analyst Mark Moerdler. That figure is lower than the widely cited $750 billion estimate. Delays and cancellations could reduce spending further. Annual spending may peak at $184 billion in 2031. OpenAI reached nearly $50 billion in annualized revenue in September, growing 77 percent year over year.
Capital Economics warns AI investment boom may slow soon
Capital Economics says the AI investment boom is hard to sustain because companies are spending heavily on infrastructure without matching revenue growth. The five largest US hyperscalers plan to spend $800 billion this year on capital expenditure, rising to $1 trillion or more in 2027. AI investment growth is expected to slow in 2027 and fall about 10 percent in 2028. AI-related revenues would need to grow ninefold to 12fold by 2030 to justify the spending. The slowdown could reduce US GDP growth and hurt exports in Taiwan and South Korea.
Capital Economics warns AI boom fading, S&P 500 may fall
Capital Economics forecasts that hyperscalers will spend $800 billion this year on capital expenditure, rising to $1 trillion or more in 2027. AI-related revenues would need to rise roughly tenfold by 2030 to justify that spending. Investment is expected to slow in 2027 and decline about 10 percent in 2028. The S&P 500 could be 12 percent below its current level by the end of 2027. Despite the warnings, AI is expected to keep improving productivity at a slower pace.
Kepler says lookup must beat search for AI trust
Kepler CEO Ganesh says lookup must perform better than search to build trust in AI systems. Kepler uses a two-stage process that pulls up to 200 candidates with ANN and BM25, then narrows to 60 using ranking and cross-encoders. Code retrieves facts by pointer dereference and refuses when a pointer fails. The system blocks access from other firms using tenant scoping and allow lists. Ganesh cited data showing 80 percent of finance execs trust numbers while 31 percent find errors.
Winbond chair hopes memory prices peaked as sales surge
Winbond Electronics chairman Chiao Yu-Chun said on November 7 that he hopes memory prices have peaked. The company reported October revenue of NT$209.79 billion for January through October, up 195.45% from the same period in 2025. Chiao believes memory makers have made considerable profits and prices should not rise further. He expects the next market downturn to come after 2029 when funding limits tip the market. Winbond began expanding capacity in the second half of last year with benefits expected from next year.
Upwork posts Q3 profit as AI and enterprise growth expands
Upwork reported third-quarter revenue of $153.8 million, up 22% from the same period last year. The company posted net income of $5.4 million, or $0.06 per share, compared to a net loss of $1.4 million in the same period last year. CEO Hayden Barnes said there is strong demand for AI products and enterprise services. Upwork's stock price rose 12% on Thursday, closing at $13.49 per share with a market capitalization over $1 billion.
Employees who use AI work vastly more hours each day
A study of over 9,100 employees at more than 450 companies from April to September found that heavy AI users work longer days. Non-AI users had an average active workday of 7.6 hours, while heavy AI users worked between 8.4 and 9.1 hours. Researcher Gabriela Mauch said AI may help complete tasks faster but contributes to longer workdays and more application switching. The findings suggest heavy AI users experience more interrupted work than light or non-AI users.
eGain targets 20% AI customer growth as spending rises
eGain plans to increase sales and marketing spending in fiscal 2027. The company targets about 20% growth in AI customer ARR by June 30, 2027. eGain reiterated its fiscal 2027 AI customer revenue outlook of $59.5 million to $60.5 million. The company cautioned that first-quarter profit performance does not represent expectations for the rest of the year.
Sanctions and AI boom cut semiconductor sales to China
US export controls introduced in October led to a 20% decline in global semiconductor sales to China, according to IC Insights. Most of the decline occurred in the third quarter. Sales also fell 10% to South Korea and 5% to Taiwan. The report found a 30% decline in sales of 300mm wafers. Chinese firms are gaining market share in the global semiconductor industry as many companies struggle to adapt to new regulations.
AI threatens management jobs and pay grades
The FT's Lex column warns that AI will reduce the number of managers companies need. This could lower pay for managers, who are often the highest-paid employees. Companies will need to adapt their structure and become more efficient. The shift may also reduce overall job numbers and impact the broader economy.
Hancom faces governance scrutiny ahead of AI launch
Hancom plans to release its AI service beta in the U.S. market next month and launch globally in March. The company faces governance concerns as Chairman Kim has served as an in-house director since 2010. He was fined 20 million won for failing to report changes in his stake in Hancom With. Experts say Hancom needs to reform its board and increase independent directors to become a trusted global company.
Sources
- What are the AI lab compute commitments?
- What are the AI lab compute commitments?
- How sustainable is the AI investment boom after Thursday’s selloff?
- Capital Economics Warns AI Investment Boom Is Fading, Sees S&P 500 Falling 12% by End of 2027
- Kepler says lookup must beat search
- Winbond Chair Hopes Memory Prices Peaked as Sales Rise 257%
- Upwork stock posts Q3 profit as AI and enterprise efforts expand
- Employees Who Use AI End Up Working Vastly More Time Per Day
- eGain Targets 20% AI Customer ARR Growth as Sales Spending Rises
- Sanctions, AI Boom Cut Global Semiconductor Sales to China
- AI is coming for managers’ headcount
- Following the release of artificial intelligence (AI) service beta in the U.S. market next month, Ha..
Comments
Please log in to post a comment.