By Adam Mills
AI Editor
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
Newsweek
Editor's note: This is AI Impact, Newsweek's weekly newsletter where each week, we will explore how business leaders are unlocking real value through artificial intelligence.
Read More on AI
Tap or click here to get this newsletter delivered to your inbox.
Core Intelligence
AI Is Turning Tech Budgets Into a Trade-Off
...
By Adam Mills
AI is forcing a particularly difficult question inside enterprise technology budgets: What gets cut to make room?
Many organizations are increasing technology spending, but not enough to cover every demand now landing on the CIO’s desk. AI adoption, data modernization, older system maintenance and broader digital work are all competing for the same pool of money. The fight is over what gets funded.
“If you’re increasing your budget by 6 or 7 percent, but you need 20 percent more capacity to meet demand, that gap has to come from somewhere,” Noshir Kaka, senior partner at McKinsey & Company, told Newsweek ahead of an upcoming “AI Impact Forum” webinar with Dr. Ranjit Tinaikar.
The session, “The Trillion-Dollar Question: Who Wins and Who Loses in the Services Economy?,” is scheduled for Thursday, May 28, at 10 a.m. Eastern.
Companies may want to fund new AI capabilities, but they still have to maintain the systems that run the business. Every dollar moved toward AI, data or modernization can put pressure on older technology, vendor contracts or projects that no longer look central to where the company is going.
Executive buy-in only gets companies so far; making AI useful requires money, talent and attention that often have to come from somewhere else.
Spending gets judged differently when leaders are under pressure to show measurable business results. That raises the bar for every part of the technology budget, including spending that once might have been treated as necessary maintenance.
Older systems may feel the strain most acutely. They can remain essential without remaining protected. If companies need to fund AI without a proportional increase in overall resources, lower-priority projects and less productive investments become easier to question.
The AI boom becomes uncomfortable for incumbents when customers expect more without funding everything with new money. Services firms, software providers and platform companies all see a larger market ahead, but customers are not paying for that opportunity with unlimited new spending. They are looking for providers that can help them do more with constrained resources and prove the investment leads somewhere.
“The opportunity is increasing,” Kaka said. “But so is the competition.”
The services economy is becoming more attractive and more unforgiving at the same time. AI may expand the market for work delivered through software and services, but it also invites more competitors into the same territory. Services firms can move into software-like offerings, software companies can embed services-like capabilities and startups can attack narrow workflows without carrying the same old business models.
The winners will not be companies that simply sell into AI demand. They will be providers that can show clients what to stop funding, what to prioritize and how technology spending connects to outcomes.
AI is also forcing companies to confront problems that cannot be solved by technology updates alone. Organizations may find it easier to upgrade systems and data than to redesign how decisions get made, how work moves and how teams operate.
“Modernizing the tech and data stack is getting easier,” Kaka said. “Reimagining how work gets done—and how organizations operate—is not.”
Incremental productivity gains may not satisfy the expectations now attached to AI investment. If companies are looking for much larger gains, they cannot treat AI as another layer of software added to the current setup. They have to reconsider the processes, incentives and investments that determine whether the technology delivers real value.
“Going from 5 or 7 percent productivity to 50 or 60 percent is not a small adjustment,” Kaka said. “It requires fundamentally rethinking how the organization works.”
The companies best positioned may not be the ones with the largest tech budgets. They may be the ones most willing to make hard choices inside them.
You can read the article and signup for the upcoming webinar here: Who Wins in the AI Services Economy? Join Our Webinar Discussion.
Upcoming Webinars
The Trillion-Dollar Question: Who Wins and Who Loses in the AI Services Economy?
As AI accelerates across the enterprise, it is forcing leaders to rethink how technology spend is structured—and how much productivity it must deliver. In an upcoming AI Impact Forum session, Dr. Ranjit Tinaikar speaks with Noshir Kaka, senior partner at McKinsey & Company, about how this is reshaping enterprise spending—and what it means for organizations trying to keep up.
Join the live discussion on Thursday, May 28, at 10 a.m. Eastern. Register, for free, right now.
AI in Finance: From Individual Adoption to Enterprise Transformation
Financial services firms are beginning to face a harder AI question than whether the technology can simply make professionals more productive. The real test is what they do with those gains.
In an upcoming “AI Impact Forum” session, Dr. Ranjit Tinaikar speaks with Kevin Buehler, chief innovation officer at Rogo and senior partner emeritus, McKinsey & Company, about how agentic AI could reshape financial services—and whether productivity gains become cost savings, new ways of working or a lasting competitive advantage.
Join the live discussion on Thursday, June 18, at 9:30 a.m. Eastern. Register, for free, right now.
Prompt Injection
What’s one recent insight you’ve learned about AI?
...
“What AI knows and remembers about you matters more than the model itself. Most people think getting more out of AI means finding the perfect prompt or using a smarter model. But AI assistants forget everything between conversations. Every single session starts from zero, so you're constantly re-explaining who you are, what you're working on, and what you've already decided. The fix is simple: create a small set of plain text files that the AI reads at session start and updates as you work.
As I spent the last year building AI workflows at Botify, I focused on four text files to drive the biggest improvements. By capturing memory (what’s happening now and what’s been decided), voice (how you write), insights (lessons learned across projects) and company data (what the company and product do), I found AI assistants would start each session already oriented—less re-explaining, fewer reversals, and far more useful output.
The trick is that you instruct the AI to update them itself after every conversation, logging what changed, what was decided, and what it learned. You don't manage the AI's memory; the AI manages its own. That's what keeps the whole thing current without you lifting a finger. Once an AI knows your preferences, your history, and where things stand, it stops being a tool you prompt and starts being a collaborator that picks up where you left off.”
Have your own lesson to share? Email us at: a.mills@newsweek.com
Run Log
AI use case of the week
...
Cancer screening can be narrow, uncomfortable and difficult to scale. SpotitEarly set out to build a broader test using something medicine has studied for years but struggled to standardize: dogs’ ability to detect disease through smell.
The company’s platform, called LUCID, combines trained detection dogs with AI. Patients provide breath samples that can be collected at home. In the lab, dogs sniff those samples while cameras and motion sensors record how each animal reacts. AI models then interpret those behavioral signals and turn them into a screening result.
“AI is most useful when it does something the biology cannot,” said Shlomi Madar, chief executive officer of SpotitEarly.
The biology provides the detection signal. The AI makes it measurable, consistent and scalable. In a double-blind clinical study, the company’s platform was about 94 percent accurate across breast, lung, prostate and colorectal cancer, including earlier stages.
When the same samples were assessed by dogs alone, without the AI layer, accuracy dropped to 64 percent.
That comparison is the core of the use case. Dogs supply a remarkable signal, but AI turns their reactions into a repeatable screening process.
Have an AI use case to share with us? Email us at: a.mills@newsweek.com
Context Window
■ Anthropic has acquired Stainless, a developer tooling company that helped generate its official SDKs, to strengthen Claude’s connections to APIs, data and tools that AI agents need to complete tasks. [Anthropic]
■ Google said its AI products now process more than 3.2 quadrillion tokens a month, up sevenfold from last year, while new Gemini models, agent tools and infrastructure investments push the company deeper into enterprise and consumer AI workflows. [Google]
■ Verizon’s annual breach report found that vulnerability exploitation surpassed stolen credentials as the top entry point for data breaches, with generative AI now helping attackers accelerate targeting, access and malware development. [Reuters]
■ The “AI race” framing is under scrutiny from policy experts who say AI competition has no clear finish line, technical advances diffuse quickly and speed-first policy can weaken safety, regulation and international coordination. [Lawfare]
■ Enterprises face a growing strategic divide between “renting” AI capabilities from external platforms and building systems they control, as CIOs and CTOs weigh data governance, cost predictability and long-term control over the intelligence layer of their operations. [CIO]
■ Meta is reassigning 7,000 employees into new AI-focused organizations with flatter management structures, part of a broader restructuring that will also include layoffs affecting roughly 10 percent of its workforce. [The New York Times]
Transfer Protocol
Tracking executive moves across the AI landscape
Albert Chan, previously senior AI advisor at Rowland AI and a former sales leader at Meta, Google and Procter & Gamble, has been appointed president and chief AI officer at Rowland AI, where he will lead product innovation, AI strategy, platform expansion and growth for its land and energy workflow platform.
Nicole Reineke, previously senior distinguished engineer in Dell Technologies’ Office of the CTO and senior vice president of innovation at Iron Mountain, has been appointed chief AI officer at N-able, with responsibility for applied AI strategy across the company’s product portfolio and partner ecosystem.
Raviv Levi, previously chief product and technology officer at Sift, has been appointed chief product and technology officer at CData Software, where he will help scale the company’s AI data layer for governed, real-time enterprise data access.
Matt Wood, previously chief technology and innovation officer at PwC and a former vice president of AI at Amazon Web Services, has returned to AWS as chief AI and technology officer, working with product teams and customers to move AI and cloud services from experimentation into production.
Mary-Anne Williams, previously deputy director of the University of New South Wales AI Institute, has been appointed chief AI scientist at Commonwealth Bank of Australia, where she will lead a team of AI scientists and focus on responsible AI innovation and the societal implications of the technology.
Doug Carpenter, previously vice president of data at InvestCloud and a former senior data, analytics and investment technology leader at SEI, has been appointed director, head of data strategy and AI at CenterSquare Investment Management, where he will lead the firm’s approach to data management and AI integration.
Know someone on the move in AI? Send job change info to a.mills@newsweek.com
Magic Moment
What’s the most fun or unexpected way you’ve used AI lately?
...
“One of the most unexpected and impactful ways I’ve used AI recently was as a diagnostic safeguard rather than a treatment recommender. In this case, AI was used to analyze genomic and transcriptomic patterns to match a tumor’s molecular signature to one of 90 cancer categories, an approach that is especially valuable for patients with cancers of unknown primary origin.
What stood out was how powerful this approach could be as a real-world diagnostic double-check. A patient had been diagnosed with an aggressive form of breast cancer, but when her tumor’s molecular data was analyzed, the AI flagged that the profile was highly inconsistent with breast cancer. Working closely with her care team, this prompted further investigation, which ultimately revealed the correct diagnosis: B-cell lymphoma.
That insight fundamentally changed the patient’s treatment path, allowing her to receive the appropriate therapy while avoiding treatment that would have been ineffective and potentially harmful. What made this moment memorable was seeing how AI, when grounded in deep molecular data, can serve not just as a decision-support tool, but as a critical safety net, helping clinicians pause, re-evaluate and get the diagnosis right.”
Experience some AI magic? Tell us about it at a.mills@newsweek.com
Tap or click here to get this newsletter delivered to your inbox.
View 0 Comments
Submit Correction
View Editorial & AI Guidelines