The strategy stories enterprise IT leaders actually read.
Volume 343, Issue 2 · 20 articles from 6 issues
The final part of a CIO.com series argues AI-era CIOs need an 'AND' mindset that reconciles opposing demands into something new.
In the final installment of a CIO.com series, the author argues that management's hardest questions involve two conflicting demands that are both right, such as accelerating digital transformation without neglecting security.
Rather than an 'OR' mindset that chooses A or B, CIOs need an 'AND' mindset that combines both strengths into a new 'C' on a higher plane.
Because CIOs stand at the boundary between business and technology, the article outlines ten types of ambidextrous leadership they need in the AI era.
Forward-deployed engineers win pilots, a CIO argues, but can't own the cross-functional trade-offs that turn AI into cash.
The author notes that OpenAI, Anthropic, Microsoft and AWS are pouring billions into forward-deployed engineering units after MIT found that 95% of enterprise AI pilots showed no profit impact.
That model is built to convert pilots to production within a quarter, not to diffuse AI through the business.
Using a demand-forecasting example, the author shows that accuracy gains free no cash unless they change decisions on safety stock, supplier commitments or build schedules.
Outside engineers deliver local, siloed wins, while the results CFOs care about cross functions and show up in cash quarters later.
Prevention alone no longer works; CISOs must measure whether the business can keep operating when defenses fail.
Decades of cybersecurity strategy have been built around preventing attacks, but ransomware crews operating like multinationals, capable nation-state actors, and supply chain compromises mean even mature programs get breached.
The author argues the important question is no longer whether an organization can stop every attack but whether it can keep operating when defenses fail.
Cyber resilience extends past technical controls into business continuity, crisis management, operational recovery, and executive decision-making. Boards increasingly want operational impact, financial exposure, regulatory implications, and recovery capability - pushing CISOs to evolve from technology leaders into business risk leaders.
TNW ranking of 10 third-party risk management platforms on AI-driven vendor review and audit-ready evidence.
This sponsored TNW article notes that companies remain responsible when vendors holding their data are breached, and that regulators now check.
EU DORA has required financial firms to register every technology supplier contract since January 2025, U.S. banking regulators issued joint vendor guidance in 2023, and SOC 2 and ISO 27001 auditors ask which vendors hold data, while spreadsheets break down beyond a handful of vendors.
The list ranks 10 third-party risk management platforms by how much review work their AI removes, such as reading security reports and flagging weak answers, and whether results count as audit evidence.
Gartner urges CISOs to lead on AI safety, treat frontier AI as insider risk, replace recognition-based identity and fund preemptive security.
Gartner outlines priority actions for security leaders in Q4 2026 as AI-augmented attacks, deepfakes and quantum risks increase uncertainty, saying CISOs should establish their authority on AI safety with a focus on how models interact with harnesses and infrastructure.
Gartner also recommends treating frontier AI deployments as insider risks and governing agents by action privileges, since 54% of organizations have no defined way to limit agent access.
Other recommendations are to stop relying on recognition alone to authorize decisions because of deepfakes, and to budget for preemptive capabilities as AI speeds up vulnerability discovery.
AI pilots stall at scale because of data hoarding, risk gridlock and misaligned metrics, not technology.
Dia Adams argues that enterprise AI rarely fails for technical reasons, and the first problem is that business units hoard data in silos with messy permissions, so AI indexing repositories can expose payroll or strategy files that obscure folder paths used to hide.
CIOs should enforce role-based access at the retrieval layer and make business units data stewards rather than waiting for multi-year cleanups.
Second, relations between IT and legal and risk teams have become adversarial, stalling approvals, and third, success metrics are misaligned with business outcomes.
Addressing these structural bottlenecks is the way out of pilot paralysis.
Standard Chartered's Brian O'Neill says transformation is an operating model, and 'you cannot transform what you cannot see.'
At the FutureCIO Conference in Singapore, Standard Chartered's global head of group transformation Brian O'Neill described transformation as an ongoing operating model rather than an initiative.
He named four essentials: investing in the right areas, ensuring investments deliver value, monitoring organizational performance and knowing where to intervene.
Because siloed data leaves large organizations without visibility, he argues business architecture is a useful blueprint only when linked to performance data showing how the organization actually works.
CIOs share the year-end priorities they're racing to finish, from keeping promises to aligning IT with next year's plans.
With Q4 underway, CIOs describe the wins they want before year end.
Adler Advisors' Robert Field says people remember what IT promised but didn't deliver, and that budgets, tax credits and performance reviews tie projects to the calendar year.
Kenco Group's Pal Narayanan wants to prove that new automation and robotics hold up through peak logistics season, with teams and vendors ready to scale.
Other advice includes using Q4 to strengthen business alignment, gather feedback from peers and show measurable value to secure funding for next year's projects.
Crist Kolder finds 10.7% of large U.S. companies now have a chief AI officer, led by healthcare and financial services.
A Crist Kolder Associates analysis of 665 Fortune 500 and S&P 500 companies found 71, or 10.7%, now have a chief AI officer or equivalent, at firms including Boeing, Eli Lilly and McDonald's.
Healthcare has the highest concentration at 18.5%, followed by financial services at 15.3% and technology at 12.6%, and the firm projects that nearly 20% of companies will have such a role by 2027.
Co-managing partner Josh Crist says the role oversees AI implementation across the business and provides internal checks and balances, which become more important as agentic AI spreads.
EDB argues data sovereignty is becoming a board-level currency in 2027, with Postgres as the underlying platform.
In EDB-sponsored partner content on VentureBeat, Max Romanenko argues that sovereignty has shifted from an economic advantage to a necessity driven by regulation, geopolitics, continuity, privacy, security and supplier dependence.
An EDB survey of 300 executives in the U.S., Germany and India found 72% say sovereignty matters more than a year ago, but only 30% have governance built in.
Romanenko predicts Europe will outspend North America on sovereign cloud in 2027, a chief sovereignty officer role will emerge, and Postgres will be the platform underneath sovereign AI.
Vendors' AI subsidies are ending, and as context memory moves to flash, token costs inherit volatile NAND pricing.
This opinion piece cites Harvard Business Review's warning of an AI 'cost shock' as vendors stop absorbing GPU, inference and token costs.
Token volumes are exploding, with Google citing 3.2 quadrillion a month, and agentic workflows push input tokens to the top of the cost stack.
NVIDIA's CMX platform offloads the KV cache to NVMe SSDs, so inference costs now depend on flash prices, which by one index rose 472% for 30TB TLC drives in a year.
The author advises reserving flash for hot context data and placing checkpoints and other retained data on cheaper tiers.
BCG says CIOs should act as chief integration officers, tying spreading AI spend to measurable business value.
As AI spending spreads beyond IT, BCG's Vlad Lukic says CIOs should act as chief integration officers who set architecture and controls and connect spending to value instead of managing disconnected projects.
BCG's research projects that agentic systems will account for two-fifths of AI value by 2030, but autonomy alone does not create value, and the companies seeing the most return have structured operating models, data, ownership and controls around their agents.
BCG recommends allocating 10% of transformation effort to algorithms, 20% to technology and data, and 70% to people, organization and processes.
Cushman & Wakefield's CDIO and CISO explain why the operating model matters more than the CISO reporting line.
At the Boston Leadership Exchange, Cushman & Wakefield CDIO Salumeh Companieh and CISO Erik Hart argued that access and operating model matter more than who the CISO reports to; Hart presents to the board himself.
Across 53,000 employees in about 60 countries, every new application or architecture change gets a cyber risk review, a process that took 18 months to become culture.
The team measures success with business metrics such as lower cyber insurance costs and faster RFP responses, invests in storytelling training, and designs controls around how employees actually work.
Security makes risk transparent, while the business owns the decision.
A Revenera study finds vendors and CIOs both struggle to tie AI spending to outcomes as pricing models shift.
As some software providers move toward token or outcome-based pricing, many CIOs struggle to predict costs and link AI spend to results.
Revenera's study found subscription pricing still leads revenue for 66% of providers, and only 7% reported selling fewer subscriptions because of AI, though nearly a quarter expect wider agent adoption to reduce demand for seats soon.
Revenera's Segerer says some rushed AI features are underused, so CIOs should track usage and use renewals to negotiate.
He also notes that each vendor defines consumption pricing differently.
CIOs should measure AI by the business KPIs executives already track, not by adoption, to prove the value of their AI strategy.
CIO.com notes that only half of organizations move more than half of their AI projects from pilot to production, per OutSystems' State of AI Development 2026 report, and usage often fails to translate into business outcomes.
Executives including Verint CTO Rob Scudiere argue that too many firms treat adoption itself as the metric.
The article lays out five metrics tied to the KPIs leaders already track and urges CIOs to align upfront with CMOs, CHROs, CFOs and business practitioners on the 'so what' of each AI initiative, so value discussions rest on shared data.
Six trends shaping IT/OT convergence in 2026, from CIOs owning OT and OT data use to security and team collaboration.
IT and operational technology, long separated by air gaps, keep converging as IoT, analytics and AI demand OT data and connectivity.
TechTarget outlines trends for 2026: more CIOs across more industries taking OT into their remit, a need for strategic alignment between IT and OT roadmaps, rising demand for OT data despite serious data quality problems, and heightened OT security concerns as air gaps disappear and legacy systems can't be patched.
Organizations are also blending IT's cybersecurity skills with OT's reliability expertise, with input from Protiviti, DXC, Info-Tech, IDC and the University of St. Thomas.
Cheaper distilled and open-weight models are undercutting massive AI compute spending; talent and culture now matter more.
The playbook of hoarding GPUs and hiring every AI expert is failing as the demand curve for frontier compute bends.
The top five tech giants are projected to spend over $750 billion on AI infrastructure in 2026, yet Chinese labs like DeepSeek and Moonshot use distillation and open-source foundations to close the gap at a fraction of the cost.
Open-weight models reaching about 90% of frontier performance on specialized tasks weaken the case for costly APIs, leaving some firms renting out excess capacity.
The author argues AI leadership depends on engineering culture, enterprise-wide adoption and retaining elite talent.
A CIO retrained 100+ engineers through hackathons and built an AI governance portal instead of restrictive rules.
This diginomica network piece profiles an anonymized CIO who rebuilt software development around AI with a people-first approach, retraining more than 100 engineers through hackathons and workshops after publicly committing to workforce retention to ease job-security fears.
Using Claude and Devin, the team compressed a projected 18-month modernization project into four months.
Rather than restrictive policy, the CIO built an AI governance portal cataloging approved tools and educating staff, arguing employee-centered change management drove both adoption and results.
Enterprises should let business needs drive AI adoption instead of deploying it because executives mandate it.
Ross Kelly reports that Appian's Roger Lee is warning enterprises against deploying AI simply because leadership mandates it, urging companies to identify specific problems before adopting AI as a solution.
Recommendations include distinguishing AI agents from simpler rules-based automation for straightforward yes/no decisions, maintaining human oversight in regulated industries, and prioritizing 'business pull rather than technology push.'
The piece cites Accenture telling employees to curb unnecessary AI usage over rising costs, and Gartner research showing AI is often deployed poorly matched to actual operational needs.
A consultant recommends IT own data engineering, the business own analytics, and a small bridging team keep the two aligned.
Drawing on client work, Arturo Torres Arpi Acero argues that putting analytics entirely in IT or entirely in the business fails in predictable ways.
IT-owned analytics tends to have strong security and governance but slow iteration, as analysts wait in ticket queues, while business-owned analytics moves faster but produces inconsistent metrics, duplicated dashboards and declining trust in the numbers.
His recommended model splits ownership: IT runs data engineering and secure access to source systems, business units own analysis, and a small central team maintains shared definitions and standards between them.