AgenticOps and the race to control enterprise AI

Corporate IA infrastructure expenses should reach $ 309 billion By 2032. The winners will not be determined by whom has the best models; This will come back to who controls the infrastructure layer which makes the AI ​​operational on a large scale.

Safety sellers make the most aggressive movements. Palo Alto networks,, Cowsterrike And Cisco Each report of security income focused on AI increasing 70 to 80% in annual sliding while traditional sales of infrastructure decrease. The model is clear: security becomes the control plan for the AI ​​company.

“The complexity of AI’s workloads is to support the existing infrastructure to its breakdown,” notes Ali Ghodsi, CEO of Databricks, notes in a blog article. “Businesses need fundamentally new approaches to manage LIG on large scale.”

The evidence rises. According to IDC73% of companies cite the insufficient infrastructure as the main obstacle to the adoption of AI. Meanwhile, armaments opponents AI faster than businesses can deploy defenses. The infrastructure wars began.

The agentics emerges as the new battlefield

Ageticops is not the vision of a seller. It is an industry recognition that traditional IT operations cannot manage AI agents operating at the speed of the machine with human authorizations. Cisco launched the category at Cisco Live 2025but AI orchestration of Microsoft,, Google model operations and startups like Weight and prejudice are already underway to own it. The battle lines are traced.

The technical requirements are brutal. Companies deploy 50,000 AI agents need an infrastructure that manages access to interior data, real-time governance and multi-team collaboration. Traditional tools break at 5,000 agents. Mathematics do not work.

“For the very first time, security becomes an accelerator for adoption, rather than an obstacle,” said Jeetu Patel, president of Cisco and CPO, in Venturebeat in a recent interview. The change is fundamental: the security teams now allow the deployment of AI rather than blocking it.

Three pillars define business quality agents: access to unified data in all areas, collaborative environments where Netops and Secops teams work together and specially designed models that govern the actions of agents. Forrester Research confirms multi-domain visibility as critical. Sellers who master the three components will be those to be dominated. But most are struggling to deliver the same effectively.

The death of the security of the perimeter

Traditional firewalls cannot protect the workloads of AI. The evidence is overwhelming. Cloud Prisma of Palo Alto Processes 2 billion security events per day at the time of execution. Fortinet safety fabric Connects more than 500 integration points because the defense of the perimeter failed. Infinite Works on zero frustful principles, assuming a violation of each layer.

The Extended Berkeley Package filter (EBPF) has changed the game. This Linux nucleus technology allows the application of safety without the performance of 40% of traditional approaches. The isovalent acquisition of $ 2.8 billion in Cisco validated the approach. EyelashThe open-source Isovalent project, now secures production workloads at Netflix, Adobe and Capital One. The 15,000 GitHub stars reflect the adoption of businesses, not on the interest of developers.

Craig Connors, Vice-President of Cisco and CTO security, has framed the change in a recent interview with Venturebeat: “The security policy is now applied to each layer, from the silicon workload.” The involvement is clear. Security becomes an integral part of the infrastructure, not a superposition.

The material acceleration seals the transformation. Safety in silicon phase operates at Nanosecond latency. The calculation is brutal: the security defined by software adds 50-200 milliseconds. Material security adds 50 to 200 nanoseconds. It is an improvement of a million times. Sellers without silicon capacities cannot compete.

The 72 -hour operating window

The opponents armed vulnerabilities in 72 hours. Patch companies in 45 days. This difference generates 84% ​​of successful violations. Each security supplier rushes to close it.

Crowdstrike falcon prevents Blocks exploited before the existence of fixes. VMDR qualification Provides vulnerability management in real time. Tanium plot promises an automated response less than hours. Cisco’s live protection applies shields at the nucleus in a few minutes.

The economy is undeniable. Research on the Ponémon Institute shows that each hour of delayed correction costs $ 84,000 at risk of violation. Automated platforms offer a return on investment (king) in 4.7 months. Cisos cannot ignore mathematics.

“Time is all cybersecurity,” said Shlomo Kramer, CEO of Cato Networks. “Automation is not only a question of efficiency; it is a question of surviving the attacks to which human teams cannot respond quickly enough.”

Observability wars intensifies

The acquisition of Splunk of 28 billion dollars signals a greater truth: observability determines which wins the battle of the infrastructure of the AI. DatadAnd Processes 18 billions of events per day. New relic Monitoring 10 billion transactions per minute. Dynamic Follows 2.5 million cloud applications.

The issues are existential. Companies that deploy AI without observability fly blind. “You cannot secure what you cannot see,” says Etay Maor, principal director of security strategy at Cato Networks. “Observability is not optional, it is the very foundation of a secure digital transformation.”

The generative user interface represents the following border. Instead of dashboards, AI creates interfaces in real time according to the exact problem solved. ServiceNow, Splunk and emerging players like Observable bet that dynamic interfaces replace static dashboards within 24 months.

Market consolidation accelerates

The infrastructure giants assemble their armies by acquisition. Cisco paid $ 28 billion for Splunk. Palo Alto acquired Cider Security, Dig Security and Talon for a billion dollars combined. Crowdsstrike bought Restify, Humio and Preemp. The acquisition of VMware of $ 69 billion from Broadcom reshapes the entire landscape.

The speed of the platform now determines survival. Unified architectures have reduced the development time of years to months. This took 18 months to deploy now launches in 8 weeks. The engineering teams vote with their feet, joining companies that ship at a start -up speed at the company’s scale.

The AI ​​infrastructure market is expected to consolidate more than 200 suppliers at less than 20 platforms within 36 months. Gartner predicts 60% of current suppliers will not exist by 2027. The message is brutal: control the full battery or not be relevant.

The bottom line

Ageticops represents the most important architectural change since the advent of Cloud Computing. Companies that build an IA infrastructure by supposing a continuous compromise, infinite identities and machine speed attacks will prosper. Those who cling to the defense of the perimeter and the response to human speed will join Blockbuster and Kodak in the digital cemetery.

Sellers who resolve this challenge – whether Cisco, Palo Alto, Microsoft or emerging players control the next decade of corporate technology. The race is on. The clock turns. The winners are already emerging.