EPAM Systems rolled out a new AI workflow service aimed at helping enterprises test AI agents before they go live. The offering draws on EPAM's experience with complex business workflows and positions the company alongside partners like Anthropic and OpenAI. Investors are waiting to see whether larger AI projects translate into meaningful revenue starting in 2027.
Kyndryl secured major AI deals with Toyota Finance Corporation and WPP, both adopting Kyndryl's Agentic AI Framework to modernize IT operations and automate workflows. These are production deployments rather than pilot projects, which could help Kyndryl compete for higher-value, longer-term work. Analysts project Kyndryl could reach $15.2 billion in revenue and $472.5 million in earnings by 2029.
DeepSeek launched its V4.1-Flash-1017 model, which promises faster agentic execution and improved benchmark scores, including 42.3 on Terminal-Bench 4.0 and 77.0 on DeepSWE v1.1. The model introduces KV Cache Compression technology and a new Python encoding toolkit while dropping the Jinja chat template. It is available through local paths like Transformers and vLLM, though it has not yet been deployed through Inference Provider. The license supports commercial self-hosting under MIT.
SpaceX AI announced plans to rebrand as SpaceXSI, according to a post from Elon Musk on X. The rename coincides with xAI's Colossus cluster in Memphis, Tennessee, which is set to house roughly 1.44 million Nvidia GPUs by late December 2026. Colossus 1 currently holds about 230,000 GPUs including H100 and H200 chips, while Colossus 2 is expected to add GB200 and GB300 Blackwell generation chips. The project involves tens of billions in financing and a custom silicon effort called Terafab, though SpaceX has not officially confirmed the rename.
Concerns are growing that the current AI investment boom may not deliver the returns investors expect. One Wall Street Journal columnist drew parallels to past overinvestment cycles like railways in the 1840s and fiber-optic networks in 2000. The Brookings Institution estimated US AI investment could reach $9 trillion by 2032, yet some AI companies have seen share prices fall. Nvidia's P/E ratio dropped from 23 to 15 despite a 15% rise in its share price, and CoreWeave's share price fell 24% in a single week.
In other developments, the state government in Patina plans to require AI training for teachers at state universities and colleges, focusing on classroom use, lecture planning, and exam preparation while emphasizing safeguards for accuracy, data privacy, and human oversight. Consumer Reports investigated Uber's AI pricing and found possible unfair differences based on personal data, along with rising company take rates and potentially fake advertised discounts. Portal26 launched the Managed Intelligence Platform to help managed service providers gain visibility into AI usage and manage security risks. President Trump's Super Intelligence Force named Director of National Intelligence Jay Clayton as chair and Federal Trade Commission Chair Andrew Ferguson as vice chair, tasked with guiding the government's AI approach. A stock analysis compared Arista Networks and Vertiv Holdings as top AI infrastructure plays, noting that while both benefit from AI campus growth, their profit cushions differ significantly.
Key Takeaways
- EPAM Systems launched an AI workflow service for enterprise agent testing, partnering with Anthropic and OpenAI, with revenue impact expected from 2027.
- Kyndryl secured production AI deals with Toyota Finance and WPP, projecting $15.2 billion in revenue and $472.5 million in earnings by 2029.
- DeepSeek released V4.1-Flash-1017, a faster model with KV Cache Compression and improved benchmark scores, available under an MIT license.
- SpaceX AI plans to rebrand as SpaceXSI as xAI's Colossus cluster scales to 1.44 million Nvidia GPUs by late 2026.
- AI investment concerns mount as Nvidia's P/E ratio fell from 23 to 15 and CoreWeave's share price dropped 24% in one week.
- The Brookings Institution estimates US AI investment could reach $9 trillion by 2032 despite returns uncertainty.
- Patina state government will require AI training for teachers at state universities and colleges.
- Consumer Reports found possible unfair AI-based pricing differences in Uber rides along with rising take rates.
- Portal26 launched a governance platform for managed service providers to monitor AI usage and manage security risks.
- Trump's Super Intelligence Force is led by Jay Clayton as chair and Andrew Ferguson as vice chair, focusing on AI policy and enforcement.
EPAM Systems launches new AI workflow service for enterprise
EPAM Systems introduced a new AI-focused service that helps test AI agents before they are deployed. The offering uses the company's decades of experience with complex business workflows. It aims to help large organizations with reliability and governance needs. The launch fits with EPAM's broader AI efforts with partners like Anthropic and OpenAI. Investors will watch whether larger AI projects can boost revenue starting in 2027.
Kyndryl lands major AI deals with Toyota Finance and WPP
Toyota Finance Corporation and WPP partnered with Kyndryl Holdings to use its Agentic AI Framework. The deals will modernize IT operations and automate workflows across infrastructure and cloud systems. These are real production deployments, not just test projects. This could help Kyndryl compete for higher value, longer term services work. Analysts forecast Kyndryl could reach $15.2b in revenue and $472.5 million in earnings by 2029.
Experts warn AI investment boom may not deliver expected returns
A Wall Street Journal columnist warned that AI investment may not produce the returns investors expect. He compared it to past booms like railways in the 1840s and fiber-optic networks in 2000. The Brookings Institution estimated US AI investment could reach $9 trillion by 2032. Some AI companies have seen falling share prices recently. Nvidia's P/E ratio dropped from 23 to 15 despite a 15% share price rise. CoreWeave's share price fell 24% in one week.
DeepSeek releases faster Flash model with shorter outputs
DeepSeek released its V4.1-Flash-1017 model which claims faster agentic execution. The model scored higher on benchmarks like Terminal-Bench 4.0 at 42.3 and DeepSWE v1.1 at 77.0. It drops the Jinja chat template and includes a new Python encoding toolkit instead. The model uses KV Cache Compression technology. It is available through local paths like Transformers and vLLM but has no Inference Provider deployment yet. The license is MIT for commercial self-hosting.
SpaceX AI operation may become SpaceXSI amid massive GPU buildout
Elon Musk posted on X that SpaceXAI would change its name to SpaceXSI. The rename comes as xAI's Colossus cluster in Memphis, Tennessee is set to house about 1.44 million Nvidia GPUs by late December 2026. Colossus 1 has roughly 230,000 GPUs including H100 and H200 chips. Colossus 2 is expected to add GB200 and GB300 Blackwell generation chips. SpaceX has not confirmed the rename in any official statement. The project involves tens of billions in financing and a custom silicon effort called Terafab.
State universities to require AI training for teachers
The state government in Patna plans to make AI training mandatory for teachers at state universities and colleges. The goal is to help teachers use AI for classroom teaching, lecture planning, and exam preparation. Officials also say safeguards are needed for accuracy, data privacy, academic integrity, and human oversight.
Consumer Reports examines Uber's AI pricing for rides
Consumer Reports studied how Uber uses AI to set rideshare prices and found possible unfair differences based on personal data. Uber and Lyft say prices depend on supply and demand, but Consumer Reports questions that. The investigation also found company take rates rose over time and some advertised discounts may have been fake.
Portal26 launches AI governance platform for MSPs
Portal26 introduced the Managed Intelligence Platform to help managed service providers handle AI use among their clients. The platform supports visibility into AI usage, security risk management, and natural language queries for AI assessments. Portal26 also launched Point2 Academy to train partners in selling and managing AI governance services.
Trump names AI enforcer and Super Intelligence Force lead
Director of National Intelligence Jay Clayton became chair of President Trump's Super Intelligence Force. Federal Trade Commission Chair Andrew Ferguson was named vice chair and will lead a law enforcement and consumer protection coordination committee. The group was created to guide the government's approach to artificial intelligence.
Arista and Vertiv compared in AI infrastructure stock analysis
Arista Networks and Vertiv Holdings appear on a list of top AI infrastructure stocks. Arista provides data center networking while Vertiv focuses on infrastructure hardware. Both companies benefit from AI campus growth, but the analysis says their profit cushions differ significantly.
Sources
- Did AI Workflow Launch Just Shift EPAM Systems (EPAM) Investment Narrative?
- Did AI Client Wins Just Shift Kyndryl (KD) Investment Narrative?
- A warning has emerged that even if artificial intelligence (AI) changes the world, it may not delive..
- DeepSeek ships faster Flash with shorter outputs
- SpaceX AI Pivot Hides a 1.44M-GPU Nvidia Bet [2026]
- AI training set to be mandatory for teachers of state universities
- Here's What Consumer Reports Found Out When It Looked Into Uber's AI Rideshare Pricing
- Portal26 Unveils MSP-Focused AI Governance Platform: Exclusive
- Donald Trump’s AI enforcer
- Arista vs. Vertiv: Which AI Infrastructure Business Has the Bigger Profit Cushion?
Comments
Please log in to post a comment.