Scale AI Infrastructure Hits Hurdles, DeepSeek, Amazon, Meta

The year 2025 marks a period of significant advancements and evolving challenges in the artificial intelligence and quantum computing sectors. Investments in AI infrastructure alone reached an impressive $29.29 billion in the second quarter of 2025, while the quantum computing market now stands at $15 billion. Despite these substantial investments, which include over $200 billion from major tech companies and governments into data centers and quantum research, many organizations face hurdles, with 44% struggling to scale their infrastructure due to power and capacity issues. Concurrently, 79% feel pressure to adopt green energy solutions. Hybrid quantum-AI systems are already driving new discoveries in fields like drug development and financial modeling, yet a 3:1 talent gap persists for the 250,000 projected quantum jobs by 2030, spurring new upskilling programs. On September 4, 2025, several key developments unfolded. Blue Sage, a cloud-based digital lending technology firm, launched its new AI Sales Agent. This voice and text-enabled assistant, built into the Blue Sage Digital Lending Platform, helps loan officers streamline tasks such as creating contacts, retrieving loan information, running borrower scenarios, and sending emails. The AI Sales Agent is currently in client testing and will see a wider release later in 2025, with Carmine Cacciavillani, president and founder, noting its potential to redefine AI in mortgage workflows. In the broader tech sphere, Chinese AI startup DeepSeek provided details on its data collection and filtering processes for training AI models. DeepSeek states it uses public online sources and authorized third parties, employing automated filters and human review to address harmful content and biases. However, the company acknowledges that 'hallucinations,' where models generate incorrect information, remain an unavoidable problem, and DeepSeek has faced criticism for high hallucination rates and misinformation, leading to a drop in user visits. The company also identified privacy issues, copyright infringement, and bias as ongoing risks. Meanwhile, Anysphere's AI coding tool, Cursor, gained popularity by focusing intensely on product improvements rather than extensive marketing. Anysphere raised $900 million at a $9.9 billion valuation in June, with notable customers including Stripe, Instacart, and Shopify. Amazon is adopting Cursor internally, and Meta CEO Mark Zuckerberg uses it for coding. The impact of AI on the job market is also a growing concern. On September 4, 2025, a letter to the editor highlighted that AI replacing human workers, such as Salesforce eliminating 4,000 jobs due to artificial intelligence, raises questions about payroll taxes for Social Security and Medicare. The writer suggested that governments should consider laws requiring AI 'workers' to contribute to these social programs or that corporations using AI cover the difference. Interestingly, a new job trend shows humans are now being hired to correct errors in AI-generated content, with graphic designers and other freelancers finding work fixing AI mistakes. AI's role in public safety and health also draws attention. VA hospitals are utilizing AI weapon scanners like Evolv Express, despite independent testers finding them error-prone and less effective than advertised. The Federal Trade Commission in 2024 called Evolv's claims of 'more accurate threat detection' and '100% detection of guns and bombs' false or unproven. Experts warn these systems can create a false sense of security, especially with VA staff shortages, and the VA recently contracted for another similar system, Xtract One SmartGateway, also with unvalidated claims. On a positive note, a report on September 4, 2025, highlighted AI's potential to enhance natural disaster preparedness and response. Patrick S. Roberts, a senior political scientist for RAND, explained that machine learning models can process vast data to forecast fires, floods, and hurricanes more accurately, aiding in risk modeling, search-and-rescue, damage assessment, and emergency communications. However, Roberts cautioned against over-reliance, warning that it could lead to decisions misaligned with human values, emphasizing the need for clear expectations and safeguards. In the realm of mental health, Dr. Kelly Merrill Jr. of the University of Cincinnati raised concerns on September 4, 2025, about using AI for mental health support. While some users report feeling happier, Merrill warns these tools are not ready to replace human therapists, citing risks like privacy issues for minors and users becoming overly attached to AI companions, potentially affecting human relationships. He advocates for 'AI literacy' and urges AI companies to build in user protections and lawmakers to prioritize safety over profit in AI regulation.

Key Takeaways

• AI infrastructure investments reached $29.29 billion in Q2 2025, with the quantum computing market valued at $15 billion.
• Blue Sage launched its AI Sales Agent on September 4, 2025, a voice and text-enabled assistant for loan officers within its Digital Lending Platform, set for wider release later in 2025.
• Chinese AI startup DeepSeek detailed its data collection methods, admitting 'hallucinations' remain an unavoidable problem and facing criticism for high error rates.
• Anysphere's AI coding tool, Cursor, achieved popularity through product focus, raising $900 million at a $9.9 billion valuation, with Amazon adopting it internally and Meta CEO Mark Zuckerberg using it for coding.
• Salesforce eliminated 4,000 jobs due to AI, prompting concerns about AI's impact on payroll taxes and suggestions for AI 'workers' to contribute to social programs.
• A new job trend on September 4, 2025, shows humans, such as graphic designers, are being hired to correct errors in AI-generated content.
• VA hospitals use AI weapon scanners like Evolv Express, despite independent testers and the FTC finding their claims of accuracy false or unproven.
• AI offers significant potential for natural disaster preparedness and response, including more accurate forecasting and aid in search-and-rescue, but requires safeguards to align with human values.
• Dr. Kelly Merrill Jr. warned on September 4, 2025, that AI mental health tools are not ready to replace human therapists, citing risks like privacy issues for minors and user over-attachment.
• 44% of companies struggle to expand AI infrastructure due to power and capacity issues, while 79% face pressure to use green energy.

Blue Sage launches AI Sales Agent for lenders

Blue Sage of Englewood Cliffs, a cloud-based digital lending technology firm, launched its new AI Sales Agent on September 4, 2025. This voice and text-enabled assistant is built into the Blue Sage Digital Lending Platform. It helps loan officers by using natural language to create contacts, get loan information, run borrower scenarios, and send emails quickly. The AI Sales Agent is currently being tested with clients and will be widely available later in 2025, aiming to redefine how AI works in mortgage processes.

Blue Sage introduces AI Sales Agent for mortgages

Blue Sage Solutions, a New Jersey company, launched its new AI Sales Agent on Thursday. This tool provides loan officers with a voice and text-enabled assistant directly within the Blue Sage Digital Lending Platform. It helps streamline tasks like creating contacts, retrieving loan details, running borrower scenarios, and sending emails using natural language. The AI Sales Agent is currently in testing with some clients and will be available to more users later in 2025, aiming to make mortgage workflows smarter and faster.

Blue Sage unveils AI assistant for loan officers

Blue Sage, a leader in cloud-based digital lending technology, announced the early release of its new AI Sales Agent on September 4, 2025. This voice and text-enabled assistant is built directly into the Blue Sage Digital Lending Platform. It helps loan officers streamline their work by using natural language to create contacts, get loan information, run borrower scenarios, and send emails quickly. The AI Sales Agent is currently being tested with select clients and is expected to be widely available later in 2025. Carmine Cacciavillani, president and founder, stated this new capability redefines AI within the mortgage workflow.

AI and quantum computing drive 2025 tech transformation

The year 2025 marks a major shift as AI and quantum computing combine, changing industries and job needs globally. Investments in AI infrastructure reached $29.29 billion in Q2 2025, but 44% of companies struggle to expand due to power and capacity issues. Quantum computing is now a $15 billion market, with hybrid systems helping in drug discovery and financial modeling. A significant 3:1 talent gap exists for quantum-AI roles, leading to new upskilling programs to fill 250,000 projected quantum jobs by 2030. Investors are looking at energy-efficient IaaS providers, quantum-ready education platforms, and hybrid quantum-AI startups.

2025 sees major AI and quantum computing growth

In 2025, AI and quantum computing are coming together globally, changing industries and job markets. Investments in AI infrastructure reached $29.29 billion in Q2 2025, and the quantum market is now worth $15 billion. Major tech companies and governments are investing over $200 billion in data centers and quantum research. However, 44% of organizations face problems scaling their infrastructure, and 79% feel pressure to use green energy. Hybrid quantum-AI systems are leading to new discoveries in drug development and finance, while a 3:1 talent gap for 250,000 quantum jobs by 2030 is driving new training programs.

Humans now fix AI mistakes in new job trend

On September 4, 2025, a new job trend shows humans are being hired to correct errors in AI-generated content. Graphic designers and other freelancers, whose jobs were once threatened by AI, are now finding work fixing its mistakes. NBC News reporter Angela Yang discussed this trend and its meaning for the U.S. job market. This development suggests a shift in how AI impacts employment, creating new roles for human oversight.

AI offers new tools for disaster preparedness

On September 4, 2025, a report highlighted AI's potential to improve natural disaster preparedness and response. Patrick S. Roberts, a senior political scientist for RAND, explained that machine learning models can process vast data to forecast fires, floods, and hurricanes more accurately. AI tools like predictive analytics, generative AI, robotics, and computer vision can help with risk modeling, search-and-rescue, damage assessment, and emergency communications. While AI offers many benefits, Roberts also warned that relying too much on it could lead to decisions that do not match human values. Organizations must set clear expectations and safeguards for AI use.

VA hospitals use disputed AI weapon scanners

VA hospitals are using AI weapon scanners like Evolv Express, which independent testers say are error-prone and less effective than advertised. The Department of Veterans Affairs pays a subscription for these machines, despite the Federal Trade Commission in 2024 calling Evolv's claims of "more accurate threat detection" and "100% detection of guns and bombs" false or unproven. Experts like Jim Bueermann and Nikita Ermolaev warn that these systems, which primarily use electromagnetic sensing, can create a false sense of security, especially with VA staff shortages. The VA recently signed another contract for a similar system, Xtract One SmartGateway, also with unvalidated claims.

DeepSeek reveals AI training data and risks

Chinese AI startup DeepSeek shared details about how it collects and filters data to train its AI models. The company stated its data mainly comes from public online sources and authorized third parties, with no intent to collect personal information. DeepSeek uses automated filters to remove harmful content and human review to address biases. While DeepSeek is working to reduce "hallucinations," where its models generate incorrect information, it admits this remains an unavoidable problem. The company has faced criticism for high hallucination rates and misinformation scandals, leading to a drop in user visits. DeepSeek also identified other risks like privacy issues, copyright infringement, and bias.

Cursor AI coding tool grows through product focus

Michael Truell, the 25-year-old CEO of Anysphere, revealed that their AI coding tool, Cursor, became popular by focusing intensely on product improvements rather than marketing. After an initial social media push, the team entered "monk mode" in 2023, letting satisfied users spread the word. Anysphere raised $900 million at a $9.9 billion valuation in June, with customers including Stripe, Instacart, and Shopify. Amazon is also adopting Cursor internally, and Meta CEO Mark Zuckerberg uses it for coding. This success contrasts with other AI startups that rely heavily on influencers and AI-enabled optimization.

Ohio researcher warns of AI mental health risks

On September 4, 2025, Dr. Kelly Merrill Jr., an assistant professor at the University of Cincinnati, raised concerns about using AI for mental health support. While many people use AI for mental health and some report feeling happier, Merrill warns these tools are not ready to replace human therapists. He highlighted serious risks, including privacy issues for minors and the danger of users becoming overly attached to AI companions, which could affect their human relationships. Merrill stressed the need for "AI literacy" and urged AI companies to build in user protections, such as reminders to take breaks or seek professional help. He also called on lawmakers to prioritize safety over profit in AI regulation.

AI replacing jobs raises payroll tax concerns

On September 4, 2025, a letter to the editor highlighted concerns about AI replacing human workers and its impact on payroll taxes. For example, Salesforce eliminated 4,000 jobs due to artificial intelligence, which means fewer employees contributing to Social Security and Medicare. The writer, D. Anthony Peters from Laguna Beach, suggested the government should create laws requiring AI "workers" to pay into these social programs. He believes corporations using AI should cover the difference, ensuring that the value AI brings to employers is taxed.

Sources

• Cloud-based tech firm Blue Sage of Englewood Cliffs launches AI sales agent
• Blue Sage unveils AI sales agent to optimize mortgage processes
• Blue Sage Launches New AI Sales Agent
• Strategic Investment in AI Infrastructure and Quantum Workforce: Navigating the 2025 Transformation
• Strategic Investment in AI Infrastructure and Quantum Workforce: Navigating the 2025 Transformation
• In hiring reversal, humans are getting jobs fixing AI sloppiness
• AI’s Potential as a Natural Disaster Preparedness Tool
• VA hospitals pay for AI weapons scanners testers say are error-prone
• DeepSeek sheds light on data collection for AI training
• Forget the marketing playbook. The CEO behind the AI tool Cursor says 2 things made it the go-to for coders.
• Ohio researcher raises red flags on AI use in mental health
• Letters to the Editor: If AI is going to replace human workers, where will payroll taxes come from?