AI infrastructure is growing almost four times faster than the enterprise networks that carry it. Use this framework to decide how fast to refresh, where to invest first and how to plan for quantum along the way.

 

 

In this blog:

  1. AI is outrunning your network
  2. Start with value, then work backward
  3. Three paths to an AI-ready network
  4. How to choose your path
  5. Quantum changes the timeline
  6. Where to start

 

How fast should you rebuild your network for AI?

Most infrastructure teams are asking that question right now. The pressure comes from both sides. Move too slowly and competitors pull ahead. Move too fast and you pay for capacity before it earns a return.

This post gives you a practical way to decide. It covers where AI demand is heading, where your investment earns the most, three refresh paths from 2027 to 2035 and why quantum computing belongs in your plan today.

AI is outrunning your network

Picture a highway built for steady commuter traffic. Now a convoy of trucks arrives every hour. That is what AI is doing to enterprise networks.

5%

yearly growth in enterprise networking, the network that has to carry the load [1] [2]

19.4%

yearly growth in AI infrastructure, from $135.8 billion in 2024 to $394.5 billion in 2030 [3]

53%

of surveyed respondents expect AI to drive more network demand than cloud or big data [4]

Put simply, AI demand is growing nearly four times faster than the network that carries it (see Figure 1).

Bar chart: enterprise networking grows about 5 percent a year, AI infrastructure 19.4 percent a year

Figure 1. Annual growth, enterprise networking versus AI infrastructure. Sources: [1] [2] [3]. Definitions and periods differ by publisher, so read the gap as an order of magnitude.

That is why calendar-based refresh plans are struggling, and why the clock is already running. A 2026 Cisco report, built on research by Foundry, surveyed more than 3,400 IT and networking decision-makers in 15 countries. Of those, 73 percent said their campus and branch networks have already hit capacity limits or will within 24 months. Respondents expect traffic to roughly triple over three years, and 93 percent are speeding up modernization because of AI demand [5]. The lesson: don't wait for the limit to find you. Plan your refresh around the AI workloads you expect to run.

The good news: You can find out where you stand in a month

Run these four checks:

  1. Count devices past end-of-sale. These are your first refresh candidates.
  2. Count devices with unpatched vulnerabilities. These are your biggest exposure.
  3. Compare power draw. Measure legacy equipment against current models doing the same work.
  4. List the AI projects landing on your network in the next 24 months. Note the bandwidth and latency each one needs.

With those answers in hand, the pace decision gets much easier. Next, let's look at where your money works hardest.

 

Start with value, then work backward

Before you size a network, look at where the money is going and what it has returned so far. Gartner forecasts worldwide AI spending of $2.6 trillion in 2026, up 47 percent from 2025, and $3.5 trillion in 2027. Its forecast table puts AI infrastructure at more than half of the total [6]. IDC sees AI infrastructure spending alone climbing from $318 billion in 2025 to $497 billion in 2026 and $1.08 trillion by 2029 [7]. The two firms define the market differently, so compare the direction, not the totals (see Figure 2).

Stacked bar chart of Gartner worldwide AI spending: $1.76 trillion in 2025, $2.60 trillion in 2026 and $3.49 trillion in 2027, with AI infrastructure at 55, 55 and 54 percent of the total

Figure 2. Worldwide AI spending, AI infrastructure versus all other AI spending, 2025–2027. Computed from Gartner's forecast table; "other" is the total minus AI infrastructure. 2026 and 2027 are forecasts. Source: Gartner [6].

Spending is not the same as return

Boards want proof, and the evidence shows a gap. In an IBM survey of 2,000 CEOs, only 25 percent said their AI initiatives had delivered the ROI they expected, and only 16 percent had scaled enterprise-wide. Half said fast investment had left them with fragmented technology [8]. McKinsey's 2026 survey of 1,719 respondents points the same way. Of those, 44 percent report AI scaling across the enterprise and 37 percent attribute at least some EBIT impact to AI, but just 6 percent are high performers, with 5 percent or more of EBIT coming from AI [9]. Cisco's 2025 AI Readiness Index finds the same thin top tier: about 13 percent of organizations qualify as the most AI-ready "Pacesetters" [10] (see Figure 3).

Horizontal bar chart of AI results by survey: 44 percent scaling AI across the enterprise (McKinsey), 37 percent report any EBIT impact (McKinsey), 25 percent of initiatives delivered expected ROI (IBM), 16 percent scaled enterprise-wide (IBM), 13 percent are most AI-ready Pacesetters (Cisco), 6 percent are high performers (McKinsey)

Figure 3. Share of organizations reporting AI results, by survey. Each bar comes from a different survey with its own sample and question wording, so compare within a source, not across sources. Sources: McKinsey [9], IBM [8], Cisco [10].

 

Expect a multi-year payback

Plan on patience. Deloitte surveyed 1,854 senior executives and found that most reached a satisfactory return on a typical AI use case within two to four years. That is well beyond the seven to 12 months Deloitte describes as typical for technology investments. Only 6 percent saw a return in under a year [11] (see Figure 4). Gartner's analysts add that CIOs are struggling to prove the value of their AI investments [6]. Build the business case on the longer clock.

Column chart of how long a typical AI use case takes to return its investment: under 1 year 6 percent, 1 to 2 years 27 percent, 2 to 3 years 37 percent, 3 to 4 years 25 percent, 4 to 5 years 4 percent, 5 or more years 1 percent

Figure 4. Time for a typical AI use case to return its investment, share of respondents. Source: Deloitte survey of 1,854 senior executives in 14 countries across Europe and the Middle East, Aug.–Sep. 2025 [11]. Results may differ in other regions.

 

Where the network fits

This is where the network earns its place in the business case. In Cisco's survey of more than 8,000 AI leaders, 54 percent said their networks cannot scale for the complexity or data volume of AI, and only 15 percent called their networks flexible or adaptable. The Pacesetters look different. Ninety-eight percent are designing networks for the growth, scale and complexity of AI, against 46 percent overall, and 71 percent say their networks can scale instantly for any AI project, against 15 percent overall. And 90 percent of Pacesetters report gains in profitability, productivity and innovation, compared with about 60 percent overall [10] (see Figure 5). Deloitte adds that one in four respondents cite inadequate infrastructure and data as a barrier to ROI [11]. These findings show an association, not proof that network spending causes the gap. They do show that readiness and results travel together.

Grouped bar chart comparing Cisco Pacesetters with all respondents: networks designed for AI scale 98 versus 46 percent; networks fully flexible 71 versus 15 percent; gains in profitability, productivity and innovation 90 versus about 60 percent

Figure 5. Pacesetters versus all respondents. Source: Cisco AI Readiness Index 2025, a Cisco survey of more than 8,000 senior IT and business leaders [10]. Cisco reports the overall figure for gains as approximately 60 percent.

 

What this means for where you invest

  • Prove the use case first. With only one in four initiatives hitting expected ROI [8], name two or three applications with a measurable business result before you size the network beneath them.
  • Fund the whole journey. Plan for a payback measured in years [11]. Cisco's Pacesetters are far more likely to have both short- and long-term funding plans: 96 percent, against 43 percent overall [10].
  • Fix the network where it is the constraint. A network that cannot scale caps every AI project that depends on it. The four baseline checks above show you where.
  • Treat power and placement as design inputs. The International Energy Agency expects AI to drive surging electricity demand from data centers [12], so where you place new capacity can matter more than any single equipment choice.

The takeaway is simple. Prove the applications first. Build only what those use cases need underneath. Use cloud capacity where the numbers work.

 

Three paths to an AI-ready network

All three paths start in 2026. They differ in pace, capital and risk. The refresh percentages below are our planning ranges, not forecasts.

Line chart of cumulative infrastructure refresh for three options from 2026 to 2035

Figure 3. Cumulative refresh by option. Midpoints of planning ranges are plotted for Options 2 and 3, and Option 3 is interpolated between period-end points.

Option 1: Transformative transition (2027–2030)

This path fits organizations under intense competitive pressure. You can refresh 20 to 30 percent of your estate each year, your technical team is experienced and you accept high risk. Think technology leaders, digital-native companies and competitive service providers.

You refresh about 20 percent in 2027 while assessing the estate and running first AI pilots. By the end of 2028, you reach about 50 percent and launch the first production AI use cases. By the end of 2029, you reach about 80 percent and decide the balance between on-premises and cloud. In 2030, you optimize what you built, target 15 to 20 percent efficiency gains and commit the final 20 percent to growth and quantum preparation. The targets by 2030: payback in about three years, 10 to 15 percent lower operating cost and measurable revenue from AI.

Option 2: Balanced evolution (2027–2032)

This path fits established organizations facing moderate pressure, with balanced finances and a preference for proven technology. Mid-market enterprises, regional providers and regulated industries are typical fits.

You start with conservative pilots in 2027, refreshing 10 to 15 percent. You reach 45 to 55 percent by 2029 as deployments hit production scale, 80 to 85 percent by 2030 and 90 to 95 percent by 2032 with quantum preparation. The upside is a lower annual capital hit and the chance to learn from early adopters. The trade-off is a competitive gap we estimate at two to three years, plus higher costs to support legacy gear longer.

Option 3: Foundation building (2027–2035)

This path fits stable markets, organizations with real financial or regulatory constraints and those focused on the quantum era. Government agencies, utilities and cloud-first businesses are typical fits.

Through 2028, you invest 10 to 15 percent in total and get AI value from cloud services. From 2029 to 2031, you refresh 15 to 20 percent for critical workloads. From 2032 to 2035, you deploy quantum-ready infrastructure in one comprehensive refresh instead of two. The risks are a competitive gap we estimate at five to nine years, and assumption risk around the quantum timeline, cloud economics and market stability.

How to choose your path

Your choice comes down to a decision point on which of two futures arrives.

If preparedness wins

If urgency wins

Interim technology goes obsolete quicklyCompetitors lock in advantages
Patience avoids wasted spendDelay costs more than action
Markets move slowly enough to catch upTalent moves to more innovative employers
Quantum arrives on schedule and rewards waitingThe quantum timeline slips, leaving aging gear with no quantum benefit

Source: authors' scenario analysis.

Match the path to your situation

  • Choose Option 1 when urgency is probable: competitive markets, technology-dependent business models and strong delivery capability.
  • Choose Option 2 when pressure is moderate and neither future clearly wins.
  • Choose Option 3 when preparedness is probable: stable markets, genuine constraints and a viable cloud-first strategy.

Four facts that hold on every path

  • Modernization is no longer optional. In a 2026 Cisco survey, 93 percent of IT leaders said they are speeding up modernization because of AI demand [5].
  • Weak foundations hold AI back. Of businesses investing heavily in AI, 57 percent report innovation delays caused by foundational issues, according to a 2026 Kyndryl and Nokia paper that cites Kyndryl's own 2025 research [8].
  • Integration decides the value. AI is not a standalone investment, and outdated network architectures throttle performance and cut returns, writes a Kyndryl executive [4].
  • Gains compound. This one is our view, not a cited finding: better performance, security and resilience open capabilities well beyond efficiency.

Whichever path you pick, the next risk is one most teams haven't priced in yet.

Quantum changes the timeline

Quantum computing won't replace classical computing. It adds a specialized tier for problems like cryptography, optimization and molecular simulation. McKinsey reports that quantum computing companies earned more than $1 billion worldwide in 2025 and projects a $43 billion to $71 billion market by 2035 [9]. Its 2025 monitor put the range at $28 billion to $72 billion [10]. Forecasts are still moving. Your plan shouldn't wait for them to settle.

The urgent issue is cryptography. Quantum computers threaten RSA, elliptic-curve cryptography and Diffie–Hellman key exchange. The risk is already real, because an adversary can capture encrypted data today and decrypt it once quantum capability matures. Security teams call this "harvest now, decrypt later." NIST says it will deprecate and remove quantum-vulnerable algorithms from its standards by 2035, with high-risk systems moving sooner [14]. It finalized its transition guidance, NIST IR 8547, in October 2026 [15].

Four actions make sense now, whichever path you choose:

  • Inventory every cryptographic dependency across network, storage and applications.
  • Start testing NIST-standardized post-quantum algorithms.
  • Migrate high-value, long-retention data first.
  • Require vendors to deliver crypto-agile architectures.

Quantum will also reshape network design. Plan for high-bandwidth, low-latency interconnects between AI clusters and quantum systems. Plan for orchestration that runs hybrid pipelines, where classical compute prepares and processes quantum tasks. Build a quantum-safe fabric with cryptographic agility across your WAN and data center. And plan early for cooling and power, because quantum systems need cryogenic environments, which most enterprises will reach through colocation or hyperscale partners.

Where to start

  1. Baseline your estate. Measure end-of-sale exposure, unpatched vulnerabilities and power draw.
  2. Rank your AI use cases by return. Let the winners define the infrastructure you need.
  3. Pick your path. Match pace to your market, your finances and your team.
  4. Start the cryptography inventory now. It is low cost, and every path needs it.

The next decade rewards leaders who match pace to circumstance and decide with evidence. The question is not whether to transform. It is how and when. World Wide Technology (WWT) can help you answer it, and help you make a new world happen.

 

 

Contributors

Oswald Fernandes

CCIE x 4 (EI,SP,Voice,DC)

BSIT (Honors -Magna Cum Laude)

 

Sources:

[1]     Precedence Research, "Enterprise network infrastructure market companies, size & trends 2026–2034," Precedence Research, Sep. 16, 2026. [Online]. Available: https://www.precedenceresearch.com/enterprise-network-infrastructure-market

[2]     Grand View Research, "Enterprise networking market size, share report, 2025–2030," Grand View Research, 2025. [Online]. Available: https://www.grandviewresearch.com/industry-analysis/enterprise-networking-market

[3]     MarketsandMarkets, "AI infrastructure industry worth $394.46 billion by 2030," MarketsandMarkets Press Release, Rep. SE 7201, Dec. 2, 2024. [Online]. Available: https://www.marketsandmarkets.com/PressReleases/ai-infrastructure.asp

[4]     P. Savill, "How network modernization enables AI success and quantum readiness," TechRadar Pro, Feb. 8, 2026. [Online]. Available: https://www.techradar.com/pro/how-network-modernization-enables-ai-success-and-quantum-readiness

[5]     I. Doescher, "Most campus and branch networks aren't ready for AI traffic," Telecompetitor, Jun. 15, 2026 (reporting a Cisco report featuring research by Foundry). [Online]. Available: https://www.telecompetitor.com/most-campus-and-branch-networks-arent-ready-for-ai-traffic/

[6]     Gartner, Inc., "Gartner forecasts worldwide AI spending to grow 47% in 2026," Gartner Newsroom, May 19, 2026. [Online]. Available: https://gartner.com/en/newsroom/press-releases/2026-05-19-gartner-forecasts-worldwide-ai-spending-to-grow-47-percent-in-2026

[7]     J. P. Seminara, "AI infrastructure spending holds near $90 billion in Q1 2026 as ARM overtakes x86 in accelerated servers; 2026 forecast raised to $497 billion," IDC Blog, Jul. 21, 2026. [Online]. Available: https://www.idc.com/resource-center/blog/ai-infrastructure-spending-holds-near-90-billion-in-q1-2026-as-arm-overtakes-x86-in-accelerated-servers-2026-forecast-raised-to-497-billion/

[8]     IBM Institute for Business Value, "IBM study: CEOs double down on AI while navigating enterprise hurdles," IBM Newsroom, May 6, 2025. [Online]. Available: https://newsroom.ibm.com/2025-05-06-ibm-study-ceos-double-down-on-ai-while-navigating-enterprise-hurdles

[9]     McKinsey & Company, "The state of AI in 2026: On the road to ROI," McKinsey & Company, Aug. 25, 2026. [Online]. Available: https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

[10]   Cisco Systems, "Cisco AI research: The most AI-ready companies outpace peers in the race to value," Cisco Newsroom, Oct. 14, 2025. [Online]. Available: https://newsroom.cisco.com/c/r/newsroom/en/us/a/y2025/m10/cisco-ai-research-the-most-ai-ready-companies-outpace-peers-in-the-race-to-value.html

[11]   J. Michalski et al., "AI ROI: The paradox of rising investment and elusive returns," Deloitte, Oct. 22, 2025. [Online]. Available: https://www.deloitte.com/nl/en/issues/generative-ai/ai-roi-the-paradox-of-rising-investment-and-elusive-returns.html

[12]   International Energy Agency, "AI is set to drive surging electricity demand from data centres while offering the potential to transform how the energy sector works," IEA News, 2025. [Online]. Available: https://www.iea.org/news/ai-is-set-to-drive-surging-electricity-demand-from-data-centres-while-offering-the-potential-to-transform-how-the-energy-sector-works

[13]   Kyndryl and Nokia, "Powering secure, AI-ready enterprise connectivity," Kyndryl, 2026. [Online]. Available: https://www.kyndryl.com/content/dam/kyndrylprogram/doc/en/2026/secure-ai-ready-enterprise-connectivity.pdf

[14]   H. Soller et al., "McKinsey quantum technology monitor 2026: A commercial tipping point," McKinsey & Company, Apr. 28, 2026. [Online]. Available: https://www.mckinsey.com/capabilities/mckinsey-technology/our-insights/mckinsey-quantum-technology-monitor-2026-a-commercial-tipping-point

[15]   H. Soller, M. Gschwendtner, S. Shabani, and W. Svejstrup, "The year of quantum: From concept to reality in 2025," McKinsey & Company, Jun. 23, 2025. [Online]. Available: https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/the-year-of-quantum-from-concept-to-reality-in-2025

[16]   National Institute of Standards and Technology, "Post-quantum cryptography," NIST Computer Security Resource Center, updated Oct. 27, 2026. [Online]. Available: https://csrc.nist.gov/projects/post-quantum-cryptography

[17]   D. Moody, R. Perlner, A. Regenscheid, A. Robinson, and D. Cooper, "Transition to post-quantum cryptography standards," NIST IR 8547, Oct. 2026. doi: 10.6028/NIST.IR.8547.

[18]   National Institute of Standards and Technology, "Module-lattice-based key-encapsulation mechanism standard," FIPS 203, Aug. 2024. [Online]. Available: https://doi.org/10.6028/NIST.FIPS.203

[19]   National Institute of Standards and Technology, "Module-lattice-based digital signature standard," FIPS 204, Aug. 2024. [Online]. Available: https://doi.org/10.6028/NIST.FIPS.204

[20]   National Institute of Standards and Technology, "Stateless hash-based digital signature standard," FIPS 205, Aug. 2024. [Online]. Available: https://doi.org/10.6028/NIST.FIPS.205