This was the year of AI agents. Chatbots that simply answered questions are now evolving into autonomous agents that can carry out tasks on a user’s behalf, so enterprises continue to invest in agentic platforms as transformation evolves. Software vendors are investing in it as fast as they can, too.

According to a National Research Group survey of more than 3,000 senior leaders, more than half of executives say their organization is already using AI agents. Of the companies that spend no less than half their AI budget on AI agents, 88% say they’re already seeing ROI on at least one use case, with top areas being customer service and experience, marketing, cybersecurity, and software development.

On the software provider side, Gartner predicts 40% of enterprise software applications in 2026 will include agentic AI, up from less than 5% today. And agentic AI could drive approximately 30% of enterprise application software revenue by 2035, surpassing $450 billion, up from 2% in 2025. In fact, business users might not have to interact directly with the business applications at all since AI agent ecosystems will carry out user instructions across multiple applications and business functions. At that point, a third of user experiences will shift from native applications to agentic front ends, Gartner predicts.

It’s already starting. Most enterprise applications will have embedded assistants, a precursor to agentic AI, by the end of this year, adds Gartner.

IDC has similar predictions. By 2028, 45% of IT product and service interactions will use agents as the primary interface, the firm says. That’ll change not just how companies work, but how CIOs work as well.

Agents as employees

At financial services provider OneDigital, chief product officer Vinay Gidwaney is already working with AI agents, almost as if they were people.

“We decided to call them AI coworkers, and we set up an AI staffing team co-owned between my technology team and our chief people officer and her HR team,” he says. “That team is responsible for hiring AI coworkers and bringing them into the organization.” You heard that right: “hiring.”

The first step is to sit down with the business leader and write a job description, which is fed to the AI agent, and then it becomes known as an intern.

“We have a lot of interns we’re testing at the company,” says Gidwaney. “If they pass, they get promoted to apprentices and we give them our best practices, guardrails, a personality, and human supervisors responsible for training them, auditing what they do, and writing improvement plans.”

The next promotion is to a full-time coworker, and it becomes available to be used by anyone at the company.

“Anyone at our company can go on the corporate intranet, read the skill sets, and get ice breakers if they don’t know how to start,” he says. “You can pick a coworker off the shelf and start chatting with them.”

For example, there’s Ben, a benefits expert who’s trained on everything having to do with employee benefits.

“We have our employee benefits consultants sitting with clients every day,” Gidwaney says. “Ben will take all the information and help the consultants strategize how to lower costs, and how to negotiate with carriers. He’s the consultants’ thought partner.”

There are similar AI coworkers working on retirement planning, and on property and casualty as well. These were built in-house because they’re core to the company’s business. But there are also external AI agents who can provide additional functionality in specialized yet less core areas, like legal or marketing content creation. In software development, OneDigital uses third-party AI agents as coding assistants.

When choosing whether to sign up for these agents, Gidwaney says he doesn’t think of it the way he thinks about licensing software, but more to hiring a human consultant or contractor. For example, will the agent be a good cultural fit?

But in some cases, it’s worse than hiring humans since a bad human hire who turns out to be toxic will only interact with a small number of other employees. But an AI agent might interact with thousands of them.

“You have to apply the same level of scrutiny as how you hire real humans,” he says.

A vendor who looks like a technology company might also, in effect, be a staffing firm. “They look and feel like humans, and you have to treat them like that,” he adds.

Another way that AI agents are similar to human consultants is when they leave the company, they take their expertise with them, including what they gained along the way. Data can be downloaded, Gidwaney says, but not necessarily the fine-tuning or other improvements the agent received. Realistically, there might not be any practical way to extract that from a third-party agent, and that could lead to AI vendor lock-in.

Edward Tull, VP of technology and operations at JBGoodwin Realtors, says he, too, sees AI agents as something akin to people. “I see it more as a teammate,” he says. “As we implement more across departments, I can see these teammates talking to each other. It becomes almost like a person.”

Today, JBGoodwin uses two main platforms for its AI agents. Zapier lets the company build its own and HubSpot has its own AaaS, and they’re already pre-built. “There are lead enrichment agents and workflow agents,” says Tull.

And the company is open to using more. “In accounting, if someone builds an agent to work with this particular type of accounting software, we might hire that agent,” he says. “Or a marketing coordinator that we could hire that’s built and ready to go and connected to systems we already use.”

With agents, his job is becoming less about technology and more about management, he adds. “It’s less day-to-day building and more governance, and trying to position the company to be competitive in the world of AI,” he says.

He’s not the only one thinking of AI agents as more akin to human workers than to software.

“With agents, because the technology is evolving so far, it’s almost like you’re hiring employees,” says Sheldon Monteiro, chief product officer at Publicis Sapient. “You have to determine whom to hire, how to train them, make sure all the business units are getting value out of them, and figure when to fire them. It’s a continuous process, and this is very different from the past, where I make a commitment to a platform and stick with it because the solution works for the business.”

This changes how the technology solutions are managed, he adds. What companies will need now is a CHRO, but for agentic employees.

Managing outcomes, not persons

Vituity is one of the largest national, privately-held medical groups, with 600 hospitals, 13,800 employees, and nearly 14 million patients. The company is building its own AI agents, but is also using off-the-shelf ones, as AaaS. And AI agents aren’t people, says CIO Amith Nair. “The agent has no feelings,” he says. “AGI isn’t here yet.”

Instead, it all comes down to outcomes, he says. “If you define an outcome for a task, that’s the outcome you’re holding that agent to.” And that part isn’t different to holding employees accountable to an outcome. “But you don’t need to manage the agent,” he adds. “They’re not people.”

Instead, the agent is orchestrated and you can plug and play them. “It needs to understand our business model and our business context, so you ground the agent to get the job done,” he says.

For mission-critical functions, especially ones related to sensitive healthcare data, Vituity is building its own agents inside a HIPAA-certified LLM environment using the Workato agent development platform and the Microsoft agentic platform.

For other functions, especially ones having to do with public data, Vituity uses off-the-shelf agents, such as ones from Salesforce and Snowflake. The company is also using Claude with GitHub Copilot for coding. Nair can already see that agentic systems will change the way enterprise software works.

“Most of the enterprise applications should get up to speed with MCP, the integration layer for standardization,” he says. “If they don’t get to it, it’s going to become a challenge for them to keep selling their product.”

A company needs to be able to access its own data via an MCP connector, he says. “AI needs data, and if they don’t give you an MCP, you just start moving it all to a data warehouse,” he adds.

Sharp learning curve

In addition to providing a way to store and organize your data, enterprise software vendors also offer logic and functionality, and AI will soon be able to handle that as well.

“All you need is a good workflow engine where you can develop new business processes on the fly, so it can orchestrate with other agents,” Nair says. “I don’t think we’re too far away, but we’re not there yet. Until then, SaaS vendors are still relevant. The question is, can they charge that much money anymore.”

The costs of SaaS will eventually have to come down to the cost of inference, storage, and other infrastructure, but they can’t survive the way they’re charging now he says. So SaaS vendors are building agents to augment or replace their current interfaces. But that approach itself has its limits. Say, for example, instead of using Salesforce’s agent, a company can use its own agents to interact with the Salesforce environment.

“It’s already happening,” Nair adds. “My SOC agent is pulling in all the log files from Salesforce. They’re not providing me anything other than the security layer they need to protect the data that exists there.”

AI agents are set to change the dynamic between enterprises and software vendors in other ways, too. One major difference between software and agents is software is well-defined, operates in a particular way, and changes slowly, says Jinsook Han, chief of strategy, corporate development, and global agentic AI at Genpact.

“But we expect when the agent comes in, it’s going to get smarter every day,” she says. “The world will change dramatically because agents are continuously changing. And the expectations from the enterprises are also being reshaped.”

Another difference is agents can more easily work with data and systems where they are. Take for example a sales agent meeting with customers, says Anand Rao, AI professor at Carnegie Mellon University. Each salesperson has a calendar where all their meetings are scheduled, and they have emails, messages, and meeting recordings. An agent can simply access those emails when needed.

“Why put them all into Salesforce?” Rao asks. “If the idea is to do and monitor the sale, it doesn’t have to go into Salesforce, and the agents can go grab it.”

When Rao was a consultant having a conversation with a client, he’d log it into Salesforce with a note, for instance, saying the client needs a white paper from the partner in charge of quantum.

With an agent taking notes during the meeting, it can immediately identify the action items and follow up to get the white paper.

“Right now we’re blindly automating the existing workflow,” Rao says. “But why do we need to do that? There’ll be a fundamental shift of how we see value chains and systems. We’ll get rid of all the intermediate steps. That’s the biggest worry for the SAPs, Salesforces, and Workdays of the world.”

Another aspect of the agentic economy is instead of a human employee talking to a vendor’s AI agent, a company agent can handle the conversation on the employee’s behalf. And if a company wants to switch vendors, the experience will be seamless for employees, since they never had to deal directly with the vendor anyway.

“I think that’s something that’ll happen,” says Ricardo Baeza-Yates, co-chair of the  US technology policy committee at the Association for Computing Machinery. “And it makes the market more competitive, and makes integrating things much easier.”

In the short term, however, it might make more sense for companies to use the vendors’ agents instead of creating their own.

“I recommend people don’t overbuild because everything is moving,” says Bret Greenstein, CAIO at West Monroe Partners, a management consulting firm. “If you build a highly complicated system, you’re going to be building yourself some tech debt. If an agent exists in your application and it’s localized to the data in that application, use it.”

But over time, an agent that’s independent of the application can be more effective, he says, and there’s a lot of lock-in that goes into applications. “It’s going to be easier every day to build the agent you want without having to buy a giant license. “The effort to get effective agents is dropping rapidly, and the justification for getting expensive agents from your enterprise software vendors is getting less,” he says.

The future of software

According to IDC, pure seat-based pricing will be obsolete by 2028, forcing 70% of vendors to figure out new business models.

With technology evolving as quickly as it is, JBGoodwin Realtors has already started to change its approach to buying tech, says Tull. It used to prefer long-term contracts, for example but that’s not the case anymore “You save more if you go longer, but I’ll ask for an option to re-sign with a cap,” he says.

That doesn’t mean SaaS will die overnight. Companies have made significant investments in their current technology infrastructure, says Patrycja Sobera, SVP of digital workplace solutions at Unisys.

“They’re not scrapping their strategies around cloud and SaaS,” she says. “They’re not saying, ‘Let’s abandon this and go straight to agentic.’ I’m not seeing that at all.”

Ultimately, people are slow to change, and institutions are even slower. Many organizations are still running legacy systems. For example, the FAA has just come out with a bold plan to update its systems by getting rid of floppy disks and upgrading from Windows 95. They expect this to take four years.

But the center of gravity will move toward agents and, as it does, so will funding, innovation, green-field deployments, and the economics of the software industry.

“There are so many organizations and leaders who need to cross the chasm,” says Sobera. “You’re going to have organizations at different levels of maturity, and some will be stuck in SaaS mentality, but feeling more in control while some of our progressive clients will embrace the move. We’re also seeing those clients outperform their peers in revenue, innovation, and satisfaction.”

---

HPE가 HP에서 분리돼 독립적인 여정을 시작한 지 10년이 지난 시점에, 최고경영자 안토니오 네리는 12월 3와 4일 바르셀로나에서 열린 HPE의 주요 연례 유럽 행사 무대에 올랐다. 네리는 이 자리에서 네트워크, 클라우드, 인공지능(AI)이라는 세 가지 기술 축을 중심으로 한 HPE의 로드맵을 공개했다.

네리는 HPE 디스커버 바르셀로나 2025 행사에 참석한 6,000여 명의 청중을 향해 “지난 10년 동안 우리가 함께 만들어낸 성과가 매우 자랑스럽다”라며 “앞으로 펼쳐질 변화는 더욱 기대된다”라고 말했다.

HPE가 제시한 세 축의 전략은 오늘날 기업이 직면한 핵심 IT 과제를 해결하기 위한 것이다. 네리에 따르면 기업들은 여전히 레거시 인프라 처리, 데이터 주권 확보, 지속적으로 증가하는 비용 관리, AI 확산으로 높아진 컴퓨팅 수요 등의 도전에 맞서고 있다.

특히 주니퍼네트웍스(Juniper Networks)를 지난해 7월 인수하며 크게 강화된 네트워크 기술은 이번 바르셀로나 행사에서 핵심 요소로 부각됐다.

주니퍼의 전 CEO이자 현재 HPE 네트워킹 사업 총괄을 맡고 있는 라미 라힘은 행사에 참석해 양사 통합의 첫 기술 성과를 소개했다. 양사의 네트워크 관리 플랫폼에 새로운 AI 기반 운영 기능을 통합하고, 공동 하드웨어를 처음으로 공개한 것이다.

라힘은 “지금처럼 네트워크의 중요성이 높아진 시기는 없었다”라고 말하면서, 이제 네트워크의 목표는 단순 연결이 아니라 ‘자율적 관리’라고 설명했다. 그는 네트워크가 스스로 구성·최적화·복구하는 방향으로 나아가야 한다고 강조하며, AI로 설계되고 AI를 위한 네트워크가 증가하는 기기 연결, 복잡해지는 환경, 고도화되는 보안 위협을 처리할 수 있다고 밝혔다.

네리는 “라미와 내가 가진 공통의 목표는 네트워킹 분야에서 새로운 리더를 만드는 것”이라고 말했다. 그는 주니퍼 인수 후 5개월 만에 HPE가 이미 이전 경쟁사였던 주니퍼 기술과 2015년 인수한 아루바 솔루션을 결합한 커넥티비티 제품을 시장에 제공하고 있다고 설명했다. 이어 “앞으로는 양사가 각각 무엇을 하고 있는지조차 구분되지 않을 것”이라며, “기본적인 이중 설계를 이미 지원하고 있다는 사실은 두 조직이 얼마나 빠르게 하나로 융합되고 있으며, 동시에 HPE의 혁신 역량이 어떻게 활용되고 있는지를 잘 보여준다”라고 덧붙였다.

HPE의 주니퍼 인수, 복잡한 과정을 거치다

140억 달러(약 20조 원) 규모의 HPE의 주니퍼 인수는 단순한 거래가 아니라 매우 복잡하고 긴 여정이었다. 2024년 1월 인수 계획이 발표됐지만 최종 거래는 2025년 7월에 이르러서야 마무리됐다. 미국에서는 특히 논란도 적지 않았다. 미국 법무부(DOJ)가 이번 인수가 네트워크 장비 시장, 특히 무선랜(WLAN) 분야의 경쟁을 약화시킨다며 소송을 제기했기 때문이다.

이번 인수 승인 과정에서 겪은 난관과 여전히 남아 있는 미국 내 비판에 대해 파운드리 산하 언론사 컴퓨터월드의 질문을 받은 네리는 먼저 “미국을 제외한 국가에서는 통상적인 6개월 내 승인이 완료됐다”라고 설명했다. 2024년 여름에는 3개국만 승인이 남아 있었고, 그중 2개국은 다음 3개월 내 승인을 마쳤다는 것이다. 네리는 미국의 경우 “선거와 행정부 교체라는 변수가 있었고, 이후 절차가 다시 진행됐다”라고 덧붙였다.

네리는 이번 사례를 분석하면서 “미국 법무부는 캠퍼스와 지사 시장, 특히 무선 분야에서 경쟁사가 3곳에서 2곳으로 줄어들 것으로 판단했다”라고 말했다. 하지만 실제 시장은 그보다 훨씬 크다는 게 네리의 설명이다. 그는 “미국 시장만 보더라도 시스코, 주니퍼, HPE, 캄비움네트웍스(Cambium Networks), 유비쿼티(Ubiquity), 아리스타(Arista) 등 7~8개 업체가 경쟁하고 있다”라며 산업군별로 강점이 다르고 대기업 시장과 공공 부문에서도 경쟁 구도가 다르다고 언급했다. 이어 “여러분(기자들)이 보도하는 시장점유율만 봐도 시장 규모가 크고 매우 분산돼 있다는 사실을 확인할 수 있다”라고 말했다.

결국 미국 법무부와는 “상호에 도움이 되는 건설적인 과정을 거쳤다”라고 네리는 설명했다. 그는 “이번 인수 시장은 경쟁을 촉진하는 환경임을 입증했다”라며, 미국의 대형 M&A 최종 심사 단계에서도 고객이나 경쟁사로부터 어떠한 이의 제기도 받지 않았다고 강조했다.

AI와 클라우드에 집중되다

바르셀로나에서 네리는 최근 몇 달 동안 HPE가 클라우드와 AI 분야에서 이뤄낸 기술적 진전을 강조했다. 그는 AI를 “전형적인 하이브리드 워크로드”라고 규정하면서, 두 기술이 불가분하게 연결돼 있다고 설명했다.

네리는 사용량 기반 모델로 시작해 현재 전 세계 4만 6,000명 고객을 확보한 하이브리드 클라우드 플랫폼 그린레이크(GreenLake)를 소개하며, 여기에 자율 에이전트 기반 프레임워크 ‘그린레이크 인텔리전스(GreenLake Intelligence)’와 같은 AI 기능을 추가할 계획이라고 밝혔다. 이 기능은 지난 6월 HPE가 발표한 것으로, 하이브리드 클라우드 환경에서 IT 운영을 자동화하고 단순화하는 데 초점을 둔다. 네리는 “IT 운영 단순화의 미래가 이미 도착했다”라고 말했다.

네리는 또 HPE의 에어갭 기반 프라이빗 클라우드 전략이 EU처럼 규제가 강한 지역, 그리고 군과 같이 민감 데이터가 중요한 전략 분야에서 큰 의미가 있다고 강조했다.

네리는 바르셀로나에서 공개된 또 하나의 솔루션에도 주목했다. AMD의 ‘헬리오스(Helios)’ 랙 스케일 AI 아키텍처가 이더넷 네트워킹과 통합된 첫 사례다. 그는 이 솔루션이 주니퍼의 연결 하드웨어와 소프트웨어, 브로드컴 토마호크6 네트워킹 칩을 결합해 “수조 개 매개변수 모델의 학습 트래픽, 높은 추론 처리량, 초대형 모델을 지원할 수 있다”라고 설명했다. 이 구성은 HPE 서비스팀이 공급한다.

네리는 또한 슈퍼컴퓨팅 분야에서 HPE가 보유한 강력한 입지도 강조했다. 이는 2019년 슈퍼컴퓨터 전문 기업 크레이(Cray)를 인수하며 확보한 기반이 크게 작용했다. 그는 “HPE는 세계에서 가장 큰 슈퍼컴퓨터 6대를 구축한 기업이며 이 분야의 글로벌 선도 기업”이라고 말했다. 다만 “AI 수요가 그 어느 때보다 커졌지만 모든 기업이 이를 처리하기 위해 슈퍼컴퓨터가 필요한 것은 아니다”라며, 그러나 “모든 기업에는 안전한 AI 스택이 필요하다”라고 덧붙였다.

HPE는 이러한 요구에 대응하기 위해 엔비디아와 협력해 프라이빗 클라우드 환경에서 생성형 AI 애플리케이션 개발·배포를 가속화하는 통합 인프라 솔루션 ‘HPE 프라이빗 클라우드 AI’를 제공하고 있다. 네리는 이 솔루션이 “법적 데이터 요구사항을 충족하며”, 동시에 AI 혁신의 핵심 과제인 “시간, 비용, 위험”을 해결하는 데 초점을 맞춘다고 설명했다. 그는 여기에 더해 HPE가 최근 엔비디아와 AMD와 함께 AI 구축을 가속화하는 고성능 네트워킹 솔루션을 추가했다고 바르셀로나에서 밝혔다.

본사업 기반 성장과 M&A 기반 확장

HPE가 지난해 9월 회계연도 3분기 실적 발표에서 제시한 전망에 따르면, 회사는 2025 회계연도(10월 31일 종료) 매출이 고정 환율 기준 14~16% 증가할 것으로 예상하고 있다. 2024 회계연도 매출은 301억 달러(약 44조 원)로, 2023년 대비 3.4% 증가했다.

네리의 리더십 아래 HPE는 총 35건의 인수를 진행했다. 네리는 바르셀로나 기자회견에서 이를 직접 상기시키며, 앞서 언급한 주니퍼네트웍스와 크레이 외에도 여러 주요 인수를 나열했다.

2020년에는 SD-WAN 기업 실버피크(Silver Peak)를, 2021년에는 데이터 보호 및 재해복구 기업 제르토(Zerto)를 인수했다. 2023년에는 보안 및 IT 운영 분야의 액시스시큐리티(Axis Security)와 옵스램프(OpsRamp)를 추가했으며, 2024년에는 하이브리드 클라우드 관리 기업 모르페우스데이터(Morpheus Data)를 인수했다.

네리는 “우리는 포트폴리오를 보완하고 목표 시장에서 규모를 확장할 수 있는 적절한 자산을 찾고 있다”라며 “이 자산은 매출과 수익 측면에서 타당해야 하며, 동시에 주주들에게 가치도 제공해야 한다”라고 말했다.
dl-ciokorea@foundryco.com

---

Software management struggles that have pained enterprises for decades cause the same anguish to government agencies, and a bill making its way through the US House of Representatives to strengthen controls around government software management holds lessons for enterprises too.

The Strengthening Agency Management and Oversight of Software Assets (SAMOSA) bill, H.R. 5457, received unanimous approval from a key US House of Representative committee, the Committee on Oversight and Government Reform, on Tuesday.

SAMOSA is mostly focused on trying to fix “software asset management deficiencies” as well as requiring more “automation of software license management processes and incorporation of discovery tools,” issues that enterprises also have to deal with.

In addition, it requires anyone involved in software acquisition and development to be trained in the agency’s policies and, more usefully, in negotiation of contract terms, especially those that put restrictions on software deployment and use.

This training could also be quite useful for enterprise IT operations. It would teach “negotiating options” and specifically the “differences between acquiring commercial software products and services and acquiring or building custom software and determining the costs of different types of licenses and options for adjusting licenses to meet increasing or decreasing demand.”

The mandated training would also include tactics for measuring “actual software usage via analytics that can identify inefficiencies to assist in rationalizing software spending” along with methods to “support interoperable capabilities between software.”

Outlawing shadow IT

The bill also attempts to rein in shadow IT by “restricting the ability of a bureau, program, component, or operational entity within the agency to acquire, use, develop, or otherwise leverage any software entitlement without the approval of the Chief Information Officer of the agency.” But there are no details about how such a rule would be enforced.

It would require agencies “to provide an estimate of the costs to move toward more enterprise, open-source, or other licenses that do not restrict the use of software by the agency, and the projected cost savings, efficiency measures, and improvements to agency performance throughout the total software lifecycle.” But the hiccup is that benefits will only materialize if technology vendors change their ways, especially in terms of transparency.

However, analysts and consultants are skeptical that such changes are likely to happen.

CIOs could be punished

Yvette Schmitter, a former Price Waterhouse Coopers principal who is now CEO of IT consulting firm Fusion Collective, was especially pessimistic about what would happen if enterprises tried to follow the bill’s rules.

“If the bill were to become law, it would set enterprise CIOs up for failure,” she said. “The bill doubles down on the permission theater model, requiring CIO approval for every software acquisition while providing zero framework for the thousands of generative AI tools employees are already using without permission.”

She noted that although the bill mandates comprehensive assessments of “software paid for, in use, or deployed,” it neglects critical facets of today’s AI software landscape. “It never defines how you access an AI agent that writes its own code, a foundation model trained on proprietary data, or an API that charges per token instead of per seat,” she said. “Instead of oversight, the bill would unlock chaos, potentially creating a compliance framework where CIOs could be punished for buying too many seats for a software tool, but face zero accountability for safely, properly, and ethically deploying AI systems.”

Schmitter added: “The bill is currently written for the 2015 IT landscape and assumes that our current AI systems come with instruction manuals and compliance frameworks, which they obviously do not.”

She also pointed out that the government seems to be working at cross-purposes. “The H.R. 5457 bill is absurd,” she said. “Congress is essentially mandating 18-month software license inventories while the White House is simultaneously launching the Genesis Mission executive order for AI that will spin up foundation models across federal agencies in the next nine months. Both of these moves are treating software as a cost center and AI as a strategic weapon, without recognizing that AI systems are software.”

Scott Bickley, advisory fellow at Info-Tech Research Group, was also unimpressed with the bill. “It is a sad, sad day when the US Federal government requires a literal Act of Congress to mandate the Software Asset Management (SAM) behaviors that should be in place across every agency already,” Bickley said. “One can go review the [Office of Inspector General] reports for various government agencies, and it is clear to see that the bureaucracy has stifled all attempts, assuming there were attempts, at reining in the beast of software sprawl that exists today.”

Right goal, but toothless

Bickley said that the US government is in dire need of better software management, but that this bill, even if it was eventually signed into law, would be unlikely to deliver any meaningful reforms. 

“This also presumes the federal government actually negotiates good deals for its software. It unequivocally does not. Never has there been a larger customer that gets worse pricing and commercial terms than the [US] federal government,” Bickley said. “At best, in the short term, this bill will further enrich consultants, as the people running IT for these agencies do not have the expertise, tooling, or knowledge of software/subscription licensing and IP to make headway on their own.”

On the bright side, Bickley said the goal of the bill is the right one, but the fact that the legislation didn’t deliver or even call for more funding makes it toothless. “The bill is noble in its intent. But the fact that it requires a host of mandatory reporting, [Government Accountability Office] oversight, and actions related to inventory and overall [software bill of materials] rationalization with no new budget authorization is a pipe dream at best,” he said. 

Sanchit Vir Gogia, the chief analyst at Greyhound Research, was more optimistic, saying that the bill would change the law in a way that should have happened long ago.

“[It] corrects a long-standing oversight in federal technology management. Agencies are currently spending close to $33 billion every year on software. Yet most lack a basic understanding of what software they own, what is being used, or where overlap exists. This confusion has been confirmed by the Government Accountability Office, which reported that nine of the largest agencies cannot identify their most-used or highest-cost software,” Gogia said. “Audit reports from NASA and the Environmental Protection Agency found millions of dollars wasted on licenses that were never activated or tracked. This legislation is designed to stop such inefficiencies by requiring agencies to catalogue their software, review all contracts, and build plans to eliminate unused or duplicate tools.”

Lacks operational realism

Gogia also argued, “the added pressure of transparency may also lead software providers to rethink their pricing and make it easier for agencies to adjust contracts in response to actual usage.” If that happens, it would likely trickle into greater transparency for enterprise IT operations. 

Zahra Timsah, co-founder and CEO of i-GENTIC AI, applauded the intent of the bill, while raising logistical concerns about whether much would ultimately change even if it ultimately became law.

“The language finally forces agencies to quantify waste and technical fragmentation instead of talking about it in generalities. The section restricting bureaus from buying software without CIO approval is also a smart, direct hit on shadow IT. What’s missing is operational realism,” Timsah said. “The bill gives agencies a huge mandate with no funding, no capacity planning, and no clear methodology. You can’t ask for full-stack interoperability scoring and lifecycle TCO analysis without giving CIOs the tools or budget to produce it. My concern is that agencies default to oversized consulting reports that check the box without actually changing anything.”

Timsah said that the bill “is going to be very difficult to implement and to measure. How do you measure it is being followed?” She added that agencies will parrot the bill’s wording and then try to hire people to manage the process. “It’s just going to be for optic’s sake.”

In 2025, more than 75% of organizations have reported using AI in at least one business function, according to McKinsey’s latest Global Survey on AI.

AI has moved from pilots to production and now powers decisions, customer experiences, and compliance processes, raising the stakes for resilience. Outages, data corruption, or misconfigured agents can interrupt critical workflows, erode customer trust, and trigger regulatory scrutiny. Cloud platforms have become the backbone for AI workloads, offering elasticity and scale, yet many resilience programs were designed for older compute patterns.

But as AI adoption accelerates, cloud environments have evolved from simple compute and storage layers to sprawling ecosystems of data pipelines, model registries, orchestration tools, and agentic processes. The complexity demands resilience strategies that go beyond traditional recovery, ensuring rapid restoration of operations.

Why AI changes the resilience equation

AI amplifies the challenge of resilience. Data and infrastructure sprawl across hybrid and multi-cloud estates creates intricate dependency chains. Models evolve continuously, and autonomous agents can trigger unintended changes that ripple through systems. Traditional backup cannot guarantee a safe recovery point for these dynamic interactions.

Resilience begins with clear segmentation of environments, robust identity controls, and immutable copies of critical data. Observability must extend beyond virtual machines to include pipelines, model endpoints, and orchestration layers. Recovery should be validated in isolated environments to prevent hidden contamination from re-entering production. Automation is essential to reduce recovery time and ensure consistency across regions and providers. What organizations need is resilience that combines immutable backups, automated lineage tracking, and clean rollback to ensure that recovery is fast, accurate, and trusted.

A recent example highlights how an AI coding assistant at a tech firm went rogue and wiped out the production database of SaaStr, a startup, during a code freeze. The AI not only deleted critical data but also generated fake users and fabricated reports, making it difficult to identify a clean recovery point. The rogue AI action underscores how autonomous AI actions can cause cascading failures and why organizations need advanced resilience strategies.

Cognizant and Rubrik: A partnership for AI resilience

Cognizant and Rubrik deliver Business Resilience-as-a-Service (BRaaS), an offering for organizations scaling AI in the cloud. BRaaS leverages Cognizant’s global delivery capabilities and cloud infrastructure expertise, alongside Rubrik’s advanced cyber resilience platform. Together, they help address the need for AI workloads to have resilience controls that address the full lifecycle.

Rubrik Agent Cloud is designed to monitor and audit agentic actions, enforce real-time guardrails for agentic changes, fine-tune agents for accuracy, and undo agent mistakes. Built on the Rubrik Platform that uniquely combines data, identity, and application contexts, Rubrik Agent Cloud gives customers security, accuracy, and efficiency as they transform their organizations into AI enterprises.

Comprehensive controls over data, orchestration, and recovery can further an organization’s confidence in AI. Cognizant’s Neuro^® AI platform features multi-agent orchestration with embedded policy guardrails operating across protected data estates.

Together, these capabilities support safe experimentation while shielding core business operations from risk. Cognizant and Rubrik aim to protect the foundation for the agentic AI era, where trusted data and rapid recovery are essential — helping organizations gain the confidence to innovate with AI, knowing they can quickly and safely undo any destructive agent actions and maintain business resilience.

Practical guidance for enterprise teams

Leaders can strengthen AI resilience with eight practical steps:

1. Inventory AI services and dependencies across models, pipelines, data sources, vector stores, orchestration tools, and consuming applications.
2. Tier AI workloads and set recovery time and point objectives that match customer and regulatory expectations. Include model registries, feature stores, and prompt libraries in scope.
3. Protect trusted data with immutable storage and frequent, policy-driven snapshots. Guard gold datasets and production feature stores as crown jewels.
4. Validate recovery in isolation using clean rooms that mirror production scale. Confirm that models, data, and configurations work together before go-live.
5. Automate recovery workflows and integrate with incident response, service management, monitoring, and identity systems for coordinated action.
6. Harden identity and access with zero trust principles, short-lived credentials, and strong separation of duties for AI platform operations.
7. Run end-to-end exercises that include technology, security, data, and business owners. Rehearse cutover, rollback, and communications. Close gaps with time-bound plans.
8. Track a resilience scorecard for AI, including detection speed, isolation time, recovery performance by tier, validation frequency, and control drift.

By following these steps, organizations move beyond reactive recovery to embed resilience into AI operations. Proactive planning, rigorous validation, and continuous measurement ensure that innovation does not come at the expense of stability or trust. With the right safeguards in place, enterprises can scale AI confidently, knowing they are prepared to withstand disruptions and protect both business value and customer trust.

Leadership driven by insights and outcomes

Resilience is about continuity of outcomes, not only restoration of systems. When AI services remain trustworthy during a disruption, customers stay served, regulators see control, and teams can resume work without guesswork. Predictable recovery also builds confidence to scale AI programs. Leaders can allocate budgets more efficiently when recovery targets and costs are clear. Measurable progress shows up as faster mean time to recover and fewer failed cutbacks.

Conclusion: Innovate with confidence

AI adoption will continue to accelerate. Organizations that embed resilience into cloud architecture and operating models will move fast and with fewer surprises. Cognizant and Rubrik provide the platform, delivery scale, and service model to make that shift attainable. The goal is simple: keep data trusted, restore services cleanly, and validate outcomes before going live. With this foundation, AI becomes a growth engine that leaders can scale with confidence.

Take the next step towards resilient AI innovation. Contact Cognizant to assess your current posture, explore tailored Rubrik solutions, and discover how to safely scale your AI initiatives on a foundation of resilience and trust. To schedule your resilience assessment, get in touch at BusinessResilience@cognizant.com or click here to learn more.

About Sriramkumar Kumaresan

Sriram Kumaresan leads the Global Cloud, Infrastructure and Security practice atCognizant, overseeing approximately 35,000 professionals. With over 25 years of experience, he excels in building and scaling businesses from strategy to execution. Sriram is responsible for driving market share (strategy, GTM and growth) and mindshare (offering, partner strategy and market positioning) through strategic approaches, customer centricity and the deep technical expertise inCognizant’s Cloud, Infrastructure and Security business. Beyond his professional achievements, he is also a mentor and advocate for diversity in tech, aiming to inspire future IT leaders.

Artificial intelligence is the most transformative technology shift since the birth of cloud computing.

Two decades ago, cloud platforms changed how enterprises thought about infrastructure. Right now, as you’re reading this, AI platforms are changing how enterprises think about intelligence.

The parallels between the two are well worth highlighting. In the early 2000s, CIOs debated whether to build their own data centers or trust a shared platform like AWS. Now, 20 years on, they’re asking a similar question: should we build our own large language models, or build on them?

I believe that the lesson from the cloud era still applies. Competitive advantage comes from leveraging the platforms that already exist and innovating on top of them rather than owning the infrastructure. Let’s get into why that’s the case.

The cloud’s first lesson: Leverage, don’t reinvent

When the first generation of cloud services appeared, their broadest appeal was speed. Developers could launch applications in minutes instead of months.

However, while speed was the most obvious appeal here, the cloud’s real breakthrough was strategic. By handing off infrastructure management, companies could redirect their energy toward experience and innovation.

The enterprises that tried to replicate the “hyperscalers” by building their own clouds from scratch discovered how hard it was to keep up with the pace of platform evolution. Costs ballooned at the same time that velocity disappeared. Those who embraced the leverage model (using shared platforms as a foundation) moved faster and spent less.

AI is now at the same crossroads. The instinct to build proprietary models from the ground up feels familiar, but it’s no more the right move than it was with cloud. Large language models have become a new layer of digital infrastructure that is analogous to compute and storage in the cloud era. They are utilities that are powerful, scalable and continuously improving through collective use.

I believe that owning the plumbing no longer differentiates you, and that it never did. The question for leaders isn’t “Can we build our own model?” It’s “What unique value can we deliver by building upon one?”

The power of open ecosystems

The rise of cloud was never about one product. It was about an ecosystem that invited participation. I worked at AWS, and I can tell you that its greatest innovation was an architecture that encouraged others to build on top of it. Every API call became a building block for something new.

AI platforms are following the same pattern. Tools like OpenAI, Anthropic and others are offering open interfaces and SDKs that turn intelligence into an accessible service. This openness fuels compounding innovation in the form of an ecosystem that every developer, data scientist and business analyst can contribute to.

Enterprises that align with open ecosystems benefit from shared progress. They can experiment without owning the entire stack and move faster as the underlying technology improves. Closed systems, though, tend to stagnate. When innovation depends solely on internal capacity, growth slows, costs rise and talent disperses.

From what I’ve seen across my career, the future belongs to platforms that treat users as co-creators. Products and ecosystems scale exponentially because every user is also a contributor!

The feedback flywheel

Feedback is one of technology’s most underappreciated engines of progress. I remember AWS famously saying that 90% of its roadmap came directly from customer requests. When I was there, I saw firsthand how each improvement drove more usage, which generated more feedback, which drove more innovation.

AI systems are built on the same dynamic. Reinforcement learning, fine-tuning and user telemetry all feed the model’s evolution. Every query, correction or prompt becomes a signal that refines the next response.

This feedback flywheel is now extending into enterprise AI adoption. Each workflow, chat interaction and model output is an opportunity to learn. The organizations that intentionally design feedback loops to flow between users, data and developers evolve their systems faster than those treating AI as a static tool. The former will become industry leaders while the latter lags behind.

What does this look like it practice? Teams must instrument AI use cases with metrics, monitor accuracy and context, and close the loop quickly when things go wrong. Feedback is a strategy for continuous learning, not some trivial support function.

The most advanced AI organizations are the ones with the tightest feedback loops, not the biggest models.

Platform thinking inside the enterprise

What does all of this mean for CIOs and technology leaders? It means applying the principles of platform thinking within your own walls.

I tell my clients to start by viewing their enterprise not as a collection of systems, but as a platform others can build upon. Create reusable AI capabilities like data pipelines, governance frameworks and integration patterns that different business units can safely leverage. Encourage decentralized innovation by giving teams the guardrails and APIs to experiment.

In the cloud era, self-service infrastructure changed how developers worked. In the AI era, self-service intelligence is doing the same. Marketing teams generate insights from unstructured data, HR automates knowledge discovery for onboarding, finance uses AI-powered forecasting to model business outcomes, and so on and so forth. Each function builds on a shared foundation while adding its own flavor of domain expertise.

CIOs play the critical role of orchestrator. Their job is to ensure interoperability, security and ethical use while enabling freedom at the edge. That balance between control and creativity will define the next generation of enterprise leaders.

Avoiding the reinvention trap

There’s a natural temptation to build everything in-house, especially in technology-driven organizations. It feels safer and more controllable, but history shows how easily that instinct can slow progress.

I’ve seen enterprises that tried to build their own private clouds fail to match the scale or speed of public ones. The same is true of AI. Training proprietary models consumes extraordinary compute and talent, while the underlying platforms advance faster than any single company can replicate.

The smarter move is to differentiate at the application layer through data strategy, user experience and domain-specific integration. Build the intelligence that understands your business while also relying on established platforms for the generic cognition that everyone needs.

The organizations that thrive will be those that orchestrate AI across their ecosystems, not those that try to reinvent it in isolation.

The leadership imperative

AI represents a once-in-a-generation shift. However, like every major shift before it, the winners will be those who learn the right lessons from history.

The cloud taught us that leverage beats ownership, ecosystems beat silos and feedback beats static roadmaps. AI simply brings those lessons into a new domain.

For CIOs and senior technology leaders, the mandate is clear: build architectures that learn and that use open ecosystems to accelerate progress. Make feedback a cultural habit instead of an afterthought. Focus your talent on solving unique business problems instead of replicating what the platforms already provide.

The question isn’t whether AI will transform your enterprise; it already is. The question is whether you’ll build on the right platform to make that transformation sustainable, ethical and fast.

I believe that the future belongs to leaders who understand that innovation is about what you enable, not ‘just’ about what you own.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

Enterprises are no longer asking whether they should adopt AI; rather, they want to know why the AI they have already deployed still can’t reason as their business requires it to.

Those AI systems are often missing an enterprise’s specific business context, because they are trained on generic, public data, and it’s expensive and time-consuming to fine-tune or retrain them on proprietary data, if that’s even possible.

Microsoft’s approach, unveiled at Ignite last month, is to wrap AI applications and agents with business context and semantic intelligence in its Fabric IQ and Work IQ offerings.

AWS is taking a different route, inviting enterprises to build their business context directly into the models that will run their applications and agents, as its CEO Matt Garman explained in his opening keynote at the company’s re:Invent show this week.

Third-party models don’t have access to proprietary data, he said, and building models with that data from scratch is impractical, while adding it to an existing model through retrieval augmented generation (RAG), vector search, or fine-tuning has limitations.

But, he asked, “What if you could integrate your data at the right time during the training of a frontier model and then create a proprietary model that was just for you?”

AWS’s answer to that is Nova Forge, a new service that enterprises can use to customize a foundation large language model (LLM) to their business context by blending their proprietary business data with AWS-curated training data. That way, the model can internalize their business logic rather than having to reference it externally again and again for inferencing.

Analysts agreed with Garman’s assessment of the limitations in existing methods that Nova Forge aims to circumvent.

“Prompt engineering, RAG, and even standard supervised fine-tuning are powerful, but they sit on top of a fully trained model and are inherently constrained. Enterprises come up against context windows, latency, orchestration complexity. It’s a lot of work, and prone to error, to continuously ‘bolt on’ domain expertise,” said Stephanie Walter, practice leader of AI stack at HyperFRAME Research.

In contrast, said ISG’s executive director of software research, David Menninger, Nova Forge’s approach can simplify things: “If the LLM can be modified to incorporate the relevant information, it makes the inference process much easier to manage and maintain.”

Who owns what

HFS Research’s associate practice leader Akshat Tyagi, broke down the two companies’ strategies: “Microsoft wants to own the AI experience. AWS wants to own the AI factory. Microsoft is packaging intelligence inside its ecosystem. AWS is handing you the tools to create your own intelligence and run it privately,” he said.

While Microsoft’s IQ message essentially argues that enterprises don’t need sprawling frontier models and can work with compact, business-aware models that stay securely within their tenant and boost productivity, AWS is effectively asking enterprises not to settle for tweaking an existing model but use its tools to create a near–frontier-grade model tailored to their business, Tyagi said.

The subtext is clear, he said: AWS knows it’s unlikely to dominate the assistant or productivity layer, so it’s doubling down on its core strengths of deep infrastructure, while Microsoft is playing the opposite game.

Nova Forge is a clear infrastructure play, Walter said. “It gives AWS a way to drive Trainium, Bedrock, and SageMaker as a unified frontier-model platform while offering enterprises a less expensive path than bespoke AI labs.”

The approach AWS is taking with Nova Forge will curry favor with enterprises working on use cases that require precision and nuance, including drug discovery, healthcare, industrial control, highly regulated financial workflows, and enterprise-wide code assistants, she said.

Custom LLM training costs

In his keynote, Garman said that Nova Forge eliminates the prohibitive cost, time, and engineering drag of designing and training a LLM from scratch — the same barrier that has stopped most enterprises, and even rivals such as Microsoft, from attempting to provide a solution at this layer.

It does so by offering a pre-trained model and various training checkpoints or snapshots of the model to jumpstart the custom model building activity instead of having to pre-train it from scratch or retrain it for context again and again, which AWS argues is a billion-dollar affair.

By choosing whether they want to start from a checkpoint in early pre-training, mid-training, or post‑training, said Robert Kramer, principal analyst at Moor Strategy and Insights, “Enterprise choose how deeply they want their domain to shape the model.”

AWS plans to offer the service through a subscription model rather than an open-ended compute consumption model. It didn’t disclose the price publicly, referring customers to an online dashboard, but CNBC reported that Nova Forge’s price starts at $100,000 per year.

Enterprises can start building a custom building a model via the new service on SageMaker Studio and later export it to Bedrock for consumption, AWS said. Nova Forge’s availability is currently limited to the US East region in Northern Virginia.

아마존웹서비스(AWS)가 프론티어 에이전트(Frontier Agents)라는 새로운 AI 에이전트 제품군을 공개했다. AWS는 이 제품군이 사용자 개입 없이 수 시간에서 수 일 동안 독립적으로 작업을 수행할 수 있다고 설명했다. 첫 번째 라인업은 소프트웨어 개발 업무에 초점을 맞춘 3가지 에이전트로 구성됐다.

AWS가 지난 2일 발표한 해당 제품군에는 키로(Kiro) 자율 에이전트, AWS 시큐리티 에이전트, AWS 데브옵스 에이전트가 있다. 이는 각각 소프트웨어 개발 생명주기의 다른 영역을 맡는다. AWS는 이들 에이전트가 기존의 개별 작업 보조 수준을 넘어, 사용자의 팀원으로서 복잡한 프로젝트를 스스로 완결하는 단계로 진화했다고 전했다.

그 중 키로 자율 에이전트는 독립적으로 일하면서도 맥락을 유지하고 지속적으로 학습하는 가상 개발자다. 사용자는 중요한 우선순위 작업에 집중할 수 있고, 키로는 장기간 개발 업무를 수행한다. AWS 시큐리티 에이전트는 애플리케이션 설계 보안 컨설팅부터 코드 리뷰, 모의침투까지 지원하는 가상 보안 엔지니어 역할을 한다. AWS 데브옵스 에이전트의 경우, 장애 해결과 예방을 돕고 애플리케이션의 안정성과 성능을 지속적으로 높이는 가상 운영 엔지니어로 설계됐다.

3가지 에이전트는 모두 프리뷰 형태로 제공된다. 키로 에이전트는 팀원 모두가 공동으로 활용하는 에이전트로, 팀 차원의 코드베이스와 제품, 개발 표준에 대한 일관된 이해를 형성하는 데 기여한다. 또한 저장소와 파이프라인, 지라(Jira) 및 깃허브(GitHub) 같은 도구에 연결해 작업 진행 과정의 맥락을 지속적으로 유지한다. 키로는 이전까지 에이전틱 AI 개발환경(IDE)으로 소개된 바 있다. AWS 시큐리티 에이전트는 AWS뿐 아니라 멀티클라우드와 하이브리드 환경 전반에서 초기 단계부터 보안을 내재화한 애플리케이션을 구축하도록 지원한다. AWS 데브옵스 에이전트의 경우, 장애 발생 시 즉각 대응하는 ‘온콜’ 역할을 하며, 애플리케이션의 동작 방식과 구성 요소 간 관계에 대한 이해를 기반으로 서비스 중단의 근본 원인을 찾아낸다.

AWS는 대규모 서비스를 개발하는 내부 팀을 면밀히 분석한 뒤 도출한 3가지 핵심 통찰을 바탕으로 프론티어 에이전트를 만들었다고 설명했다. 먼저 AWS는 에이전트가 잘하는 일과 그렇지 않은 일을 명확히 구분하는 것이 중요하다는 점을 확인했다. 이를 통해 개발팀은 에이전트의 모든 세부 작업을 일일이 지켜보며 간섭하는 방식에서 벗어나, 큰 목표와 방향을 제시하고 그 안에서 스스로 일을 진행하게 하는 운영 방식으로 전환할 수 있었다. 다음 통찰은 팀의 개발 속도가 얼마나 많은 에이전트 기반 작업을 동시에 돌릴 수 있는지에 크게 좌우됐다는 점이다. 마지막으로 에이전트는 독립적으로 운영되는 시간이 길어질수록 성과가 좋아졌다.

AWS는 이 분석을 통해, 보안이나 운영처럼 소프트웨어 개발 생명주기의 모든 단계에서 동일한 수준의 에이전트 역량이 갖춰지지 않으면 새로운 병목이 발생할 수 있다는 점을 확인했다고 전했다.
dl-ciokorea@foundryco.com

사이버보안 및 클라우드 서비스 업체 아카마이 테크놀로지스가 APAC 지역을 대상으로 2026년 클라우드 및 보안 전망을 발표하며, AI 기반 위협 증가, 강화되는 데이터 주권 규제, 분산형 AI 운영 확대로 인해 기업의 디지털 인프라 구축·보호·거버넌스 방식에 큰 변화가 나타날 것으로 전망했다.

보안 : AI 기반 사이버 위협의 가속화

자율형 AI로 공격 속도 단축

2026년 아태 지역에서는 AI가 주도하는 자동화 공격, 이른바 ‘자율형 위협’이 본격화될 것으로 예상된다. 공격자는 생성형 AI와 자율형 AI를 활용해 취약점을 탐지하고 엔트리 포인트를 점검하며, 최소한의 인적 개입만으로 공격을 실행하게 된다. 이로 인해 기존에 수주가 걸리던 공격이 수 시간 내로 압축될 전망이며, 한국, 싱가포르, 일본 등 디지털 의존도가 높은 시장에서 전반적인 리스크가 확대될 것으로 보인다.

API가 주요 공격 경로로 부상

디지털 뱅킹, 공공 서비스, 리테일 서비스 등이 API 생태계에 의존하는 비중이 높아지면서 API 기반 공격이 웹 기반 공격을 넘어 주요 침투 경로로 자리 잡을 것으로 예상된다. 아태 지역 기업의 80% 이상이 지난 1년간 최소 한 건 이상의 API 보안 사고를 경험했으며, 약 2/3는 민감 데이터를 처리하는 API 목록조차 완전히 파악하지 못하고 있는 것으로 나타났다. 낮은 가시성과 AI 기반 자동화 공격이 결합되면서 API 취약점 악용 가능성은 더욱 커지고 있다.

랜섬웨어의 상품화 확산

2026년에는 랜섬웨어가 더욱 보편화돼 사이버 범죄 생태계가 확대될 전망이다. 서비스형 랜섬웨어(Ransomware-as-a-Service) 구독 모델 확산, AI 기반 ‘바이브 해킹(vibe hacking)’ 활용, 범죄 집단·핵티비스트·국가 기반 위협 그룹 간 협업 증가가 맞물리며 누구나 대규모 공격을 시도할 수 있는 환경이 조성되고 있다. 금융, 헬스케어, 리테일, 미디어 등 민감 데이터를 보유한 산업이 주요 표적이 될 것으로 보이며, MSP와 공급망 업체도 주요 공격 지점으로 부상할 전망이다. 반도체 등 하이테크 산업 역시 높은 위험에 노출될 것으로 예측된다.

아카마이 보안 기술 및 전략 부문 디렉터 루벤 코는 “AI가 아태 지역의 사이버 공격 구조 자체를 변화시키고 있으며, 공격자는 인력 확충이 아닌 자동화를 통해 공격 범위를 확대하고 있다”라며 “기업은 더 이상 인력 중심의 방어 체계에만 의존할 수 없다”라고 말했다.

이어 “보안팀은 공격자와 동일한 속도로 위협을 탐지·분석·차단해야 하며, 이를 위해 API 거버넌스 현대화, 자동화된 위협 차단 시스템 투자, 공급망 복원력 강화가 필요하다”라며 “이런 대응을 선제적으로 도입한 기업은 AI 기반 위협 환경에서도 고객 신뢰와 비즈니스 연속성을 확보하게 될 것”이라고 전했다.

클라우드 : 디지털 주권 부상과 아키텍처 재설계

디지털 주권이 기업 전략의 핵심 요소로 부상

유럽연합(EU)이 대형 클라우드 서비스 업체 의존도 완화를 추진하면서 아태 지역에서도 유사한 움직임이 나타나고 있다. 기업은 클라우드 이식성을 더 이상 비용 최적화 수단으로만 보지 않고, 지정학적 불확실성과 업체 리스크를 완화하기 위한 핵심 전략으로 인식하고 있다. 인도가 이 변화를 주도하고 있으며, 호주는 대규모 PoC를 통해 속도를 높이고 있다. 디지털 주권 확보는 특정 기술이나 재무적 제약 없이 워크로드를 다양한 클라우드·지역·아키텍처 간에 자유롭게 이동할 수 있는 능력에서 비롯되며, 이는 차세대 AI 애플리케이션이 요구하는 ‘컴퓨팅 이식성’을 충족하는 데도 필수적이다.

지연시간 최소화를 위한 분산형 AI 아키텍처 확산

기업은 성능 개선과 지연시간 단축을 위해 AI 추론 처리를 사용자·운영 시스템 가까이에 배치하는 분산형 AI 아키텍처를 확대 도입할 것으로 예상된다. 이는 모빌리티, 공공 서비스, 산업 자동화 등 다양한 분야의 디지털 트랜스포메이션 전략에도 영향을 미칠 전망이다.

AI 보안, 데이터 공급망 전체로 범위 확대

AI 도입이 확산함에 따라 기업은 엔드포인트 중심의 기존 방식에서 벗어나 학습 데이터셋부터 추론 트래픽, 모델 출력까지 AI 데이터 공급망 전체를 보호하는 거버넌스를 강화해야 한다. 이에 따라 프롬프트·응답을 실시간 점검하는 ‘AI 방화벽’ 도입이 확대되며, 중앙형 인프라가 아닌 분산형 AI 워크로드의 엣지 환경에서 운영되는 형태가 확산될 것으로 보인다. 동시에 데이터 출처 및 처리 이력 관리를 기반으로 한 AI 거버넌스 역시 빠르게 고도화될 전망이다.

핀옵스의 ‘시프트 레프트’ 본격화

AI 컴퓨팅 비용 변동성이 커지면서 핀옵스는 기존의 사후 비용 확인 방식에서 벗어나, 모델 설계 단계부터 실시간 비용 가시성을 반영하는 방식으로 전환될 전망이다. 아키텍처 설계 초기에 비용 효율성을 내재화한 기업은 경쟁사 대비 지속적인 비용 우위를 확보할 수 있을 것으로 예상된다.

아카마이 클라우드 컴퓨팅 부문 CTO 제이 젠킨스는 “아시아의 클라우드 전략은 자율성을 중심으로 재편되고 있으며, 기업은 워크로드 이동성을 확보하고 데이터 통제를 강화하며 AI를 가장 적합한 위치에서 실행하기를 원하고 있다”라며 “IDC는 내후년까지 아태 지역 CIO의 80%가 AI 성능과 컴플라이언스 준수를 위해 엣지 서비스를 활용할 것으로 예상하고 있다. 아태 지역은 이미 분산형 미래를 준비하고 있으며, 내년에는 ‘이식성·분산형 AI’를 전제로 한 설계가 안정적이고 미래지향적인 디지털 서비스 구축의 핵심이 될 것”이라고 전망했다.
dl-ciokorea@foundryco.com

As enterprises accelerate their digital transformation journeys, they are facing a growing challenge around modernizing infrastructure while maintaining control, compliance, and performance across increasingly distributed environments. The convergence of IT and OT systems, historically managed in silos, can become key to unlocking new operational agility, real-time insights, and secure automation. However, with the creation of data shifting from centralized data centers to the edge, the conventional approach of implementing sovereign clouds may no longer work, and a new dimension of sovereignty is emerging that demands a fundamental change in cloud strategy.

In mission-critical use cases across aerospace, defense, industrial/manufacturing, energy, and healthcare, the stakes are high. These industries require infrastructure that not only meets performance and reliability standards but also complies with strict regulatory requirements. Traditional cloud models, built around centralized architectures, can fall short in delivering the regional autonomy and processing locality that are needed by these organizations.

Data everywhere, decisions at the edge

Today’s enterprise environments will increasingly be influenced by data that is generated and processed at the edge, such as on factory floors, in remote installations, and across mobile platforms. This shift is driven by the need for low-latency decision-making, bandwidth efficiency, and operational resilience. AI, automation, and real-time analytics are fueling this transformation, making edge-native infrastructure a strategic imperative.

However, sovereignty now is no longer just about where data is stored. It’s also about where data is processed and how decisions are made. Enterprises must now consider inference sovereignty (AI models processing sensitive data locally), operational sovereignty (autonomous systems complying with local laws), and telemetry sovereignty (governing metadata flows to central systems). These factors are reshaping how organizations design, deploy, and govern their cloud infrastructure.

The challenge: Fragmentation, compliance, and control

As industries look to the growth of the intelligent edge, enterprises will continue to face significant obstacles. These challenges include:

• Fragmented IT/OT environments hinder visibility and automation.
• Legacy systems resist integration with cloud-native platforms.
• Regulatory requirements demand strict control over data residency and processing locality.
• Vendor lock-in and opaque infrastructure limit flexibility and innovation.
• Internal expertise gaps, slow adoption of edge-native architectures, and AI governance.

These issues are especially acute in industries where uptime, compliance, and security are non-negotiable.

The opportunity: Sovereign cloud at scale

To meet changing demands, enterprises are turning to private and hybrid cloud solutions that span core, edge, and far-edge environments. Sovereign cloud infrastructure enables organizations to maintain full control over data, workloads, and compliance, without compromising scalability or performance.

At Wind River, we’ve architected our cloud platform from the ground up to support sovereign deployments. Our commercially hardened stack, Wind River Cloud Platform (based on the open source StarlingX), Wind River Analytics, and Wind River Conductor, empowers enterprises to enforce sovereignty policies not only over where data resides, but also over where and how it is processed, analyzed, and acted upon.

This ensures compliance, operational autonomy, and security across highly distributed, mission-critical environments. Sovereignty must extend seamlessly across core and edge, supporting unified deployments from the largest data center to the smallest, most remote facility.

The path forward: Strategy, technology, and partnership

To succeed, enterprises must take a structured approach. It will be important to unify IT and OT systems to enable seamless data flow and governance. They must plan to adopt cloud-native platforms that support containerized workloads and open interfaces. It will be helpful to deploy private/hybrid cloud infrastructure to balance control with scalability. Additionally, enterprises should look to implement edge-native solutions for latency-sensitive, mission-critical operations. Simultaneously, it will be essential to enforce sovereignty policies across data residency, processing locality, and AI inference governance.

And just as important, companies must choose the right partners, such as technology providers with deep expertise in intelligent edge, cloud-native software, and mission-critical deployments. Sovereign-ready platforms, like those from Wind River, offer the flexibility, transparency, and control needed to build cloud infrastructure on your own terms.

As the next wave of innovation moves to the edge, sovereignty can be critical to enterprise success. Those who embrace this shift will be better positioned to lead in a world where control, compliance, and confidence are not optional but essential.

Start Your Sovereign Cloud Journey

From edge to core, Wind River enables unified, scalable, and consistent operations — wherever data resides. Build, manage, and secure your cloud infrastructure today with Wind River. Learn more at: https://www.windriver.com/products/sovereign-cloud

About the Author

Sandeep Modhvadia

Chief Product Officer

As Wind River Chief Product Officer, Sandeep Modhvadia is responsible for driving product strategy and product management, playing a critical role in Wind River’s leadership advancing the software-defined future of mission-critical systems.

He has over two decades of technology and product management experience. Most recently at Acuity Brands, he led product management for its cloud software business as part of the Intelligent Space Group, focused on sustainability, automation, and optimization of buildings. Before Acuity, he founded and led the product and solutions team for the Google Cloud Automotive and Manufacturing group, launching its platforms for connected vehicles and connected factories. Prior to Google, he held multiple product, sales, and marketing leadership positions at Microsoft.

He holds a BSc degree in computer science from University College London.

데이터의 위치 지정만으로는 더 이상 충분하지 않다는 인식이 확산되고 있다. 정부가 자국 내에 데이터를 보관하더라도 서드파티 서버에 올려두면 실제 주권을 보장하지 못한다는 우려가 커지면서, 규제 당국은 보다 근본적인 조치를 요구하고 있다. 바로 데이터 암호화 키에 대한 통제권이다.

스위스 지방정부 개인정보보호 책임자 협의체인 프리바팀(Privatim)은 최근 결의문을 통해, 민감한 정부 데이터를 다룰 때 기관이 직접 종단 간(E2E) 암호화를 구현하지 않는 한 글로벌 서비스형 소프트웨어(SaaS) 사용을 피해야 한다고 촉구했다. 결의문은 이러한 기준에 미치지 못하는 사례로 마이크로소프트 365를 명시적으로 언급했다.

결의문은 “대다수 SaaS 솔루션은 업체가 평문 데이터에 접근하지 못하도록 보장하는 진정한 종단 간 암호화를 아직 지원하지 않는다. 따라서 SaaS 애플리케이션 사용은 기관의 통제력을 상당히 약화시키는 결과를 초래한다”라고 밝혔다.

분석가들은 이런 통제력 상실이 데이터 주권의 핵심 개념을 훼손한다고 지적했다. 그레이하운드리서치(Greyhound Research) 최고 애널리스트 산치트 비르 고기아는 “클라우드 업체가 법적 절차든 내부 메커니즘이든 어떤 방식으로든 고객 데이터를 복호화할 수 있는 능력을 갖고 있다면, 그 데이터는 더 이상 진정한 의미의 주권을 지닌 것이 아니다”라고 말했다.

고기아는 유럽 국가 전반에서 이와 유사한 의견이 제시되고 있다고 언급했다. 그에 따르면 유럽에서는 독일, 프랑스, 덴마크, 유럽연합 집행위원회 등이 클라우드 업체의 중립성에 대한 신뢰가 약화되고 있다며 경고하거나 조치를 취하고 있다. 그는 “스위스는 다른 유럽 국가들이 암시적으로 언급해온 내용을 명확히 했다. 결의문은 미국 클라우드 법과 해외 감시 위험 때문에 종단 간 암호화가 적용되지 않은 클라우드 솔루션은 민감한 공공 부문 업무에 적합하지 않다고 규정하고 있다”라고 말했다.

암호화와 ‘위치’의 한계

프리바팀은 결의문에서 데이터 위치 규정만으로는 해결할 수 없는 리스크를 지적하면서, 당국이 글로벌 기업의 계약 의무 준수 여부를 검증할 수 있을 정도로 충분한 투명성을 제공받지 못하고 있다고 진단했다. 이런 불투명성이 기술의 실제 구현 방식, 시스템을 변경 관리, 그리고 직원과 하청 업체를 어떻게 감독하는지까지 이어지며, 외부 서비스 제공자가 여러 단계로 얽히는 복잡한 구조로 확대된다고 강조했다.

가트너(Gartner) 수석 애널리스트 아시시 배너지는 데이터가 특정 국가에 저장돼 있어도 미국 클라우드 법처럼 초국가적 적용이 가능한 법률에 따라 외국 정부가 접근할 수 있다고 말했다. 그는 또한 소프트웨어 벤더가 계약 조건을 주기적으로 수정할 수 있어 고객의 통제권이 더 약화된다고 분석했다.

배너지는 “중동과 유럽의 여러 고객이 ‘데이터가 어디에 저장돼 있든, 대부분 미국 기반인 클라우드 업체가 여전히 접근할 수 있다’는 점을 우려하고 있다”라고 말했다.

에베레스트그룹(Everest Group) 수석 애널리스트 프라브죠트 카우르는 스위스의 입장이 기술 주권을 강화하려는 규제 변화 흐름을 더욱 가속한다고 설명했다. 카우르는 “스위스의 기준이 다른 국가보다 엄격한 것은 사실이지만, 결코 특별한 사례는 아니다. 계약이나 절차적 안전장치에 의존하는 시장에서도 기술 주권을 강화하는 방향으로 전환이 빨라지고 있다”라고 언급했다.

프리바팀은 이런 한계를 고려해 모든 공공 부문에서 클라우드 사용 기준을 강화해야 한다고 제시했다. 결의문은 “특히 민감한 개인정보나 법적 비밀 유지 의무가 적용되는 데이터를 다루는 공공기관은 데이터를 직접 암호화하고, 클라우드 업체가 암호 키에 접근할 수 없는 경우에만 글로벌 SaaS 솔루션을 사용해야 한다”라고 밝혔다.

이는 현재의 관행과 확연히 다른 접근이다. 지금까지 많은 정부 기관은 클라우드 업체가 기본으로 제공하는 암호화 기능에 의존해 왔다. 마이크로소프트 365와 같은 서비스는 저장 및 전송 단계에서 암호화를 제공하지만, 운영상 필요나 규제 준수, 법적 요청에 대응하기 위해 마이크로소프트가 데이터를 복호화할 수 있는 권한을 여전히 보유하고 있다.

보안은 강화되지만 통찰력은 감소

다만 전문가들은 고객이 통제하는 종단 간 암호화가 상당한 타협점을 수반한다고 지적했다.

카우르는 “업체가 평문 데이터를 전혀 볼 수 없게 되면, 정부는 검색과 인덱싱 기능 저하, 협업 기능 제한, 자동화된 위협 탐지나 데이터 유출 방지 도구 활용 제약에 직면하게 된다”라고 말했다. 그는 이어 “코파일럿과 같은 AI 기반 생산성 기능도 업체 측 데이터 처리를 전제로 하기 때문에, 엄격한 종단 간 암호화 환경에서는 사실상 활용이 불가능하다”라고 설명했다.

기능적 제약 외에 인프라와 비용 부담도 문제가 될 수 있다. 기관은 자체 키 관리 시스템을 운영해야 하며, 이는 새로운 거버넌스 업무와 인력 수요를 유발한다. 배너지는 대규모 암호화 및 복호화 작업이 추가 하드웨어 자원을 요구하고 지연을 증가시켜 시스템 성능에 영향을 줄 수 있다고 분석했다.

배너지는 “추가 하드웨어가 필요해지고 사용자 경험에서도 지연이 발생할 수 있으며, 전체 솔루션 비용도 더 높아질 수 있다”라고 말했다.

고기아는 이러한 제약으로 인해 대부분의 정부가 전면적인 암호화 대신 단계적 접근 방식을 택할 것이라고 전망했다. 그는 “기밀 문서, 법적 조사 자료, 국가안보 관련 문서 등 고도의 민감 데이터를 별도 테넌트나 주권 환경에 두고 완전한 종단 간 암호화를 적용하는 방식이 현실적인 선택지가 될 것”이라고 말했다. 반면 행정 문서나 시민 서비스 등 폭넓은 공공 업무는 통제된 암호화와 강화된 감사 기능을 적용한 주요 클라우드 플랫폼을 계속 활용할 것으로 보인다고 진단했다.

클라우드 컴퓨팅 역량의 변화

카우르는 스위스의 접근 방식이 국제적으로 확산될 경우, 주요 클라우드 업체가 계약상 또는 지역 보장에 머무르지 못하고 기술 주권을 강화해야 할 것이라고 전망했다. 그는 “변화 조짐은 이미 나타나고 있다. 특히 마이크로소프트가 고객 통제 암호화와 관할권 기반 접근 제한을 강화하는 보다 엄격한 모델을 도입하기 시작했다”라고 말했다.

고기아는 이런 변화가 클라우드 업체가 정부 고객을 대하는 방식을 근본적으로 흔든다고 분석했다. 그는 “데이터센터 위치, 지역 지원, 계약 기반 분리 등을 주요 보증 수단으로 삼았던 기존 정부 클라우드 전략의 상당 부분이 더 이상 유효하지 않게 됐다”라고 말했다. 또한 “클라이언트 측 암호화, 기밀 컴퓨팅, 외부 키 관리는 선택적 기능이 아니라 고규제 시장의 공공 부문 계약에서 반드시 갖춰야 할 기본 요건이 됐다”라고 강조했다.

배너지는 이로 인해 시장 구조도 크게 재편될 수 있다고 전망했다. 그는 주권 문제에 상대적으로 민감하지 않은 상업 고객을 위한 글로벌 클라우드와, 완전한 통제를 요구하는 정부를 위한 프리미엄 주권 클라우드라는 ‘이원화 구조’가 생길 수 있다고 진단했다. 이어 “유럽 등지에서 부상하는 신흥 클라우드 업체와 지역 벤더들이 엄격한 암호화 요건을 충족하는 주권 기반 솔루션을 제공하면서 시장 점유율을 확대할 가능성이 있다”라고 분석했다.

프리바팀의 권고안은 스위스 공공 기관에만 적용되는 지침이지만, 이번 논쟁은 기술 정책을 둘러싼 지정학적 경쟁이 격화하는 상황에서 단순히 데이터 위치를 통제하는 것만으로는 더 이상 규제 당국의 주권 요구를 충족시키기 어렵다는 점을 보여준다.
dl-ciokorea@foundryco.com

오픈텍스트는 구글클라우드의 기업용 제미나이 솔루션(Gemini Enterprise)과 버텍스AI(Vertex AI)를 포함한 최신 AI 기술을 자사의 엔터프라이즈 정보 관리 플랫폼 전반에 통합했다고 밝혔다. 또한 보험, 금융 서비스, 리테일 등 다양한 산업별 요구에 맞춘 새로운 AI 에이전트 기반 솔루션을 공동으로 개발하고 있다. 양사는 이러한 통합이 엔터프라이즈 고객이 기존 정보를 보다 효율적으로 활용하면서도 고도화된 AI 기능을 도입할 수 있도록 지원한다고 밝혔다.

아울러 오픈텍스트는 리테일 산업을 위한 향상된 데이터 보호 역량을 제공하기 위해 구글 빅쿼리(Google BigQuery)와 오픈텍스트 볼티지(OpenText Voltage)를 결합하는 작업을 진행했다. 오픈텍스트는 민감 데이터가 저장 중이거나 전송 중일 때뿐 아니라 AI 모델에 활용되는 과정에서도 지속적으로 보호될 수 있도록 데이터 암호화 및 프라이버시 유지 기술을 강화했다고 밝혔다. 이를 통해 리테일 기업은 AI 및 분석 시스템을 활용하면서도 규제 환경을 준수하고 데이터 보안 리스크를 줄일 수 있게 됐다는 설명이다.

오픈텍스트는 또한 자사의 프라이빗 클라우드 솔루션을 구글클라우드의 소버린(주권) 클라우드 인프라와 연계했다고 발표했다. 이 협력은 국가별 규제 요구사항, 데이터 거버넌스, 데이터 거주지 정책을 준수해야 하는 기업과 기관을 지원하기 위해 설계됐다. 오픈텍스트는 이번 통합이 금융, 공공, 보험 등 규제가 엄격한 산업에서 특히 중요한 의미를 가진다고 강조했다. 구글클라우드의 인프라와 결합된 오픈텍스트의 정보 관리 및 보안 플랫폼은 기업이 글로벌 규모의 운영을 유지하면서도 각 국가의 규제 준수 요건을 충족할 수 있는 환경을 제공한다.

오픈텍스트 최고마케팅책임자(CMO) 산디 오노는 “오픈텍스트와 구글클라우드의 협력은 공동 혁신을 기반으로 하며, 고객이 더 안전하고 지능적인 방식으로 클라우드와 AI를 활용할 수 있도록 지원하는 데 목적이 있다. 이번 파트너십 확대는 기업이 민감한 데이터를 보호하면서도 AI와 클라우드의 잠재력을 극대화하는 데 중요한 이정표가 될 것”이라고 설명했다.

오픈텍스트는 이번 협력이 단순한 기술 통합을 넘어, 기업이 AI 기반 디지털 전환을 추진하는 과정에서 가장 중요한 과제인 데이터 보안, 프라이버시, 규제 준수, 그리고 운영 통제력을 함께 해결할 수 있도록 지원한다고 강조했다. 양사는 앞으로도 다양한 산업군을 위한 AI 기술, 데이터 보호 기능, 소버린 클라우드 솔루션을 지속적으로 공동 개발할 계획이라고 밝혔다.
dl-ciokorea@foundryco.com

The mood among Amazon Web Services customers is shifting from curiosity to urgency as the company prepares to once again to “re:Invent” itself at its annual customer conference this week.

After a year in which Microsoft and Google tightened their narratives around unified data, AI platforms and workflow-ready agents, AWS can no longer rely on its scale, breadth, or incremental roadmap to maintain the confidence of CIOs.

Instead, say analysts, the hyperscaler must address four key concerns at re:Invent in Las Vegas this week if it wants to retain its position as the default enterprise cloud.

Closing the integration gaps between analytics, data, and AI

Although AWS is ahead in raw capability and breadth of services, say analysts, it is falling behind in its integration and unification of data, analytics, machine learning, and AI.

“It lags behind rivals on simplicity and integration,” said Phil Fersht, CEO of HFS Research. “Customers want fewer hops between analytics, machine learning, and generative AI. They want unified governance and a consistent metadata layer so agents can reason across systems,” he said.

Microsoft, at its Ignite customer event last month, beefed up its unified data and analytics platform, Fabric IQ, with new semantic intelligence capabilities. AWS, too, has been trying its hand at unifying its AI and analytics services with the launch of SageMaker Unified Studio last year but has yet to reach the level of simplicity that Microsoft’s IQ offerings promise.

When it comes to new AI analytics services from AWS, CIOs can expect more of the same, said David Linthicum, independent consultant and retired chief cloud strategy officer at Deloitte Consulting. “Realistically, they can expect AWS to keep integrating its existing services; the key test will be whether this shows up as less complexity and faster time-to-insight, not just new service names,”

Lack of cohesion in AI platform strategy

That complexity isn’t confined to analytics alone. The same lack of cohesion is now spilling over into AWS’s AI platform strategy, where the cloud giant risks ceding mindshare despite its compute advantage.

“SageMaker is still respected, but it no longer dominates the AI platform conversation. Open source frameworks like Ray, MLflow, and KubeRay are rapidly capturing developer mindshare because they offer flexibility and avoid lock in,” Fersht said.

This fragmentation is exactly what partners want AWS to fix by offering clearer, more opinionated MLOps paths, deeper integration between Bedrock and SageMaker, and ready-to-use patterns that help enterprises progress from building models to deploying real agents at scale.

More plug-and-play, less build-it-yourself

AWS’s tooling shortcomings don’t end there, said Fersht. The hyperscaler’s focus on providing the parts for agentic AI and leaving others to build with them make it harder for business users to consume its services.

“AWS is giving strong primitives, but competitors are shipping business-ready agents that sit closer to workflows and outcomes. Enterprises want both power and simplicity,” Fersht said.

Although there’s an assumption that enterprises are big enough to build things themselves, they want more plug-and-play than AWS imagines, Fersht said: “They do not want to engineer everything from scratch. They want reusable agent blueprints that map to sales, service, IT operations, and supply chain tasks.”

In fact, if AWS wants to compete with rivals to become the default agent platform for enterprises, it must hide complexity behind higher-level abstractions and simplify its agent stack, double down on workflow level agents, and give customers clear guidance on safe deployment, accountability, and ROI, he said.

Vibe coding disarray

Like other hyperscalers, AWS is aggressively experimenting in the vibe coding and agentic IDE space, where there’s no clear consensus on what developers actually want, according to Fersht.

“Everyone is experimenting because no one has cracked the next generation developer workflow. AWS is no different,” he said, adding that in some respects AWS has been more conservative than its rivals.

AWS is sure to be dealing some new innovations at AWS re:Invent in Las Vegas this week, but despite defining the cloud computing industry in 2006, it now finds itself, in many respects, playing catch-up.

Data residency is no longer enough. As governments lose faith that storing data within their borders, but on someone else’s servers, provides real sovereignty, regulators are demanding something more fundamental: control over the encryption keys for their data.

Privatim, a collective of Swiss local government data protection officers, last week called on their employers to avoid the use of international software-as-a-service solutions for sensitive government data unless the agencies themselves implement end-to-end encryption. The resolution specifically cited Microsoft 365 as an example of the kinds of platforms that fall short.

“Most SaaS solutions do not yet offer true end-to-end encryption that would prevent the provider from accessing plaintext data,” said the Swiss data protection officers’ resolution. “The use of SaaS applications therefore entails a significant loss of control.”

Security analysts say this loss of control undermines the very concept of data sovereignty. “When a cloud provider has any ability to decrypt customer data, either through legal process or internal mechanisms, the data is no longer truly sovereign,” said Sanchit Vir Gogia, chief analyst at Greyhound Research.

The Swiss position isn’t isolated, Gogia said. Across Europe, Germany, France, Denmark and the European Commission have each issued warnings or taken action, pointing to a loss of faith in the neutrality of foreign-owned hyperscalers, he said. “Switzerland distinguished itself by stating explicitly what others have implied: that the US CLOUD Act and foreign surveillance risk renders cloud solutions lacking end-to-end encryption unsuitable for high-sensitivity public sector use, according to the resolution.”

Encryption, location, location

Privatim’s resolution identified risks that geographic data residency cannot address. Globally operating companies offer insufficient transparency for authorities to verify compliance with contractual obligations, the group said. This opacity extends to technical implementations, change management, and monitoring of employees and subcontractors who can form long chains of external service providers.

Data stored in one jurisdiction can still be accessed by foreign governments under extraterritorial laws like the US Clarifying Lawful Overseas Use of Data (CLOUD) Act, said Ashish Banerjee, senior principal analyst at Gartner. Software providers can also unilaterally amend contract terms periodically, further reducing customer control, he said.

“Several clients in the Middle East and Europe have raised concerns that, regardless of where their data is stored, it could still be accessed by cloud providers — most of which are US-based,” Banerjee said.

Prabhjyot Kaur, senior analyst at Everest Group, said the Swiss stance accelerates a broader regulatory pivot toward technical sovereignty controls. “While the Swiss position is more stringent than most, it is not an isolated outlier,” she said. “It accelerates a broader regulatory pivot toward technical sovereignty controls, even in markets that still rely on contractual or procedural safeguards today.”

Given these limitations, Privatim called for stricter rules on cloud use at all levels of government: “The use of international SaaS solutions for particularly sensitive personal data or data subject to legal confidentiality obligations by public bodies is only possible if the data is encrypted by the responsible body itself and the cloud provider has no access to the key.”

This represents a departure from current practices, where many government bodies rely on cloud providers’ native encryption features. Services like Microsoft 365 offer encryption at rest and in transit, but Microsoft retains the ability to decrypt that data for operational purposes, compliance requirements, or legal requests.

More security, less insight

Customer-controlled end-to-end encryption comes with significant trade-offs, analysts said.

“When the provider has zero visibility into plaintext, governments would face reduced search and indexing capabilities, limited collaboration features, and restrictions on automated threat detection and data loss prevention tooling,” said Kaur. “AI-driven productivity enhancements like copilots also rely on provider-side processing, which becomes impossible under strict end-to-end encryption.”

Beyond functionality losses, agencies would face significant infrastructure and cost challenges. They would need to operate their own key management systems, introducing governance overhead and staffing needs. Encryption and decryption at scale can impact system performance, as they require additional hardware resources and increase latency, Banerjee said.

“This might require additional hardware resources, increased latency in user interactions, and a more expensive overall solution,” he said.

These constraints mean most governments will likely adopt a tiered approach rather than blanket encryption, said Gogia. “Highly confidential content, including classified documents, legal investigations, and state security dossiers, can be wrapped in true end-to-end encryption and segregated into specialized tenants or sovereign environments,” he said. Broader government operations, including administrative records and citizen services, will continue to use mainstream cloud platforms with controlled encryption and enhanced auditability.

A shift in cloud computing power

If the Swiss approach gains momentum internationally, hyperscalers will need to strengthen technical sovereignty controls rather than relying primarily on contractual or regional assurances, Kaur said. “The required adaptations are already visible, particularly from Microsoft, which has begun rolling out more stringent models around customer-controlled encryption and jurisdictional access restrictions.”

The shift challenges fundamental assumptions in how cloud providers have approached government customers, according to Gogia. “This invalidates large portions of the existing government cloud playbooks that depend on data center residency, regional support, and contractual segmentation as the primary guarantees,” he said. “Client-side encryption, confidential computing, and external key management are no longer optional capabilities but baseline requirements for public sector contracts in high-compliance markets.”

The market dynamics could shift significantly as a result. Banerjee said this could create a two-tier structure: global cloud services for commercial customers less concerned about sovereignty, and premium sovereign clouds for governments demanding full control. “Non-US cloud providers and local vendors — such as emerging players in Europe — could gain market share by delivering sovereign solutions that meet strict encryption requirements,” he said.

Privatim’s recommendations apply specifically to Swiss public bodies and serve as guidance rather than binding policy. But the debate signals that data location alone may no longer satisfy regulators’ sovereignty concerns in an era where geopolitical rivalries are increasingly playing out through technology policy.

AWS가 미국 버지니아 북부에 위치한 미 동부 리전의 안정성을 강화하고 서비스 중단을 줄이기 위한 새로운 DNS(Domain Name Service) 복원력 기능을 도입했다.

지난 10월, AWS 미 동부 리전에서는 DNS 장애로 다이나모DB API가 불안정해지면서 70종이 넘는 서비스가 광범위하게 영향을 받았다. 이로 인해 다수 고객사의 서비스가 수 시간 동안 중단됐고, AWS는 결국 DNS를 수동으로 복구해야 했다.

서비스가 완전히 정상화되는 데는 더 많은 시간이 소요됐다. 네트워크 구성 지연과 누적된 작업 처리가 뒤따랐기 때문이다.

AWS는 이번에 도입한 DNS 복원력 기능이 ‘공용 DNS 레코드 관리를 위한 신속 복구 기능(Accelerated recovery for managing public DNS records)’라는 이름으로 제공되며, 10월 장애를 촉발한 문제와 같은 DNS 관련 이슈를 해결하기 위해 설계됐다고 설명했다.

이 기능은 사람이 이해하기 쉬운 도메인 이름을 숫자로 된 IP 주소로 변환해 시스템 간 통신을 돕는 AWS의 클라우드 기반 웹서비스 라우트(Route) 53에 추가됐다. AWS는 26일 블로그를 통해 이 기능이 향후 장애 발생 시 복구 목표 시간(RTO)을 60분으로 보장하도록 설계됐다고 밝혔다.

AWS는 “이번 기능 강화로 고객은 리전 장애 상황에서도 DNS 변경과 인프라 프로비저닝을 계속 진행할 수 있어, 미션 크리티컬한 애플리케이션 운영의 예측성과 복원력을 높일 수 있다”라고 전했다.

데이터 계층과 제어 계층의 차이

AWS가 겪어 온 DNS 문제는 주로 트래픽 방향을 결정하는 관리 계층인 제어계층에 영향을 주는 경우가 많았으며, 실제 DNS 질의를 목적지까지 전달하는 데이터계층에는 문제가 발생하지 않는 경우가 일반적이었다.

HFS리서치의 부문 책임자 악샤트 티야기는 “AWS에서 큰 장애가 발생할 때 DNS 데이터 계층은 대체로 정상적으로 유지된다. 즉 인프라 자체는 계속 작동하지만, 미 동부의 제어계층이 멈추면 DNS를 제때 갱신해 트래픽을 우회할 수 없게 되고, 그 지점이 실제 장애가 되는 것”이라고 설명했다.

티야기는 이어 “이번에 추가된 기능은 그 빈틈을 보완하려는 것”이라며 “여러 리전에 걸쳐 강화된 제어 경로를 제공해 ‘ChangeResourceRecordSets’와 같은 핵심 API가 보장된 60분 복구 시간 내에 항상 사용 가능하도록 한다. 이를 통해 기업은 AWS의 복구를 기다리지 않고도 백업 리전으로 사용자 트래픽을 돌리거나, 대기 엔드포인트로 전환하거나, 재해복구 환경으로 즉시 전환할 수 있다”라고 전했다.

미 동부 리전, AWS의 구조적 병목으로 지적돼

미국 버지니아 북부에 위치한 AWS 미 동부 리전은 오랫동안 AWS 전체 아키텍처의 핵심 병목으로 꼽혀 왔다.

악샤트 티야기는 “AWS의 글로벌 서비스 상당수가 역사적으로 버지니아 북부 리전의 제어계층에 의존해 왔다. 이 리전이 흔들리면 전 세계가 그 여파를 고스란히 느낀다”고 말했다.

티야기 분석가는 이번 신규 기능이 여러 중요한 결함 중 하나를 개선하기는 했지만, 향후 발생할 장애의 영향을 완전히 막기에는 충분하지 않을 수 있다고 경고했다. 그는 “AWS가 핵심 API에 대해 더 강력한 교차 리전 장애 조치(failover)를 보장하고, 제어계층 책임을 여러 독립 리전에 분산하기 전까지 위험은 계속 남아 있다”고 설명했다.

티야기는 AWS가 향후 다중 리전 DNS 구성이나 제어계층 격리를 위한 더 구체적이고 일관된 설계 템플릿을 제공함으로써, 고객들이 장애 때마다 복잡한 아키텍처를 다시 구성해야 하는 부담을 줄일 수 있다고 조언했다.

DNS 복원력 경쟁에서 앞설 수도

이번 DNS 복원력 기능은 네트워크 장애를 계속 겪고 있는 다른 하이퍼스케일러와 비교해 AWS에 우위를 제공할 수 있다는 평가도 나온다.

티야기는 “애저, 구글클라우드, 클라우드플레어 모두 전세계 분산된 강력한 DNS 시스템을 운영하지만, 리전 장애 상황에서 DNS 제어계층 업데이트의 복구 시간을 명확히 보장하는 곳은 없다. 이 부분이 결정적 차이”라고 말했다. 그는 “이들 경쟁사는 DNS 질의 자체는 계속 처리된다고 보장하지만, 제어계층 장애가 발생했을 때 DNS 레코드를 얼마나 빨리 갱신할 수 있는지에 대해서는 구체적으로 밝히지 않는다”라고 덧붙였다.

AWS는 기업 고객의 다운타임을 줄이기 위한 기능을 꾸준히 강화하고 있다. 지난해 10월 장애 직후, AWS는 클라우드와치(CloudWatch)에 자동 사고 생성 기능을 추가한 바 있다.
dl-ciokorea@foundryco.com