From Internet Rails to AI Agents: The Story of TelCos to LLMs

Internetification to the Agenticfication of the world. The Internet Era's Playbook and Anticipating AI's Evolution Through History's Lens.

Having been deeply involved and engaged in the digital transformation journey of the world across enterprise businesses (B2B) and consumer products (B2C) over the last 20 years, I find the parallels between the era of the internetification of the world and today's AI revolution absolutely fascinating.

While chapters of books could be written about it, this is my own take in a simplified version on why this comparison offers compelling insights into what may lie ahead.

The Infrastructure Build-out: The Beginning of Every Great Wave

Late 1990s and early 2000s saw massive telco infrastructure investments that laid the foundation for global internet access.

This wasn't just about the consumer-facing telcos - behind them stood the powerhouse network equipment providers: Nokia Siemens Network, Ericsson, Alcatel-Lucent, Nortel Networks, Motorola's infrastructure division, and many others. By 2005, while global internet users reached about 1 billion+, with most access still through fixed lines in pockets of the developed economies of the world. These users were also largely business users or organizational users at work.

Today's AI landscape shows interesting similarities and notably different structure. Instead of separate infrastructure and equipment layers, we're seeing vertical integration. The cloud giants (Microsoft, Google, Amazon) are deeply intertwined with their LLM partners (OpenAI, Google's Gemini, Anthropic).

Then there's Meta, taking a fascinating different approach with their open-source Llama models - reminiscent in some ways of how certain telcos opted for open networking approaches in the earlier era. While the consumer adoption of LLMs or AI assistants has been much rapid this time, it is still the enterprises that are leading the charge to find right use cases to implement AI solutions at scale.

The Scale-up Revolution and the democratization of the internet

The iPhone's 2007 launch changed the trajectory of mobile internet access. Look at the numbers: from about 120 million smartphones in 2007 to over 1.4 billion by 2013 and multi-fold of that today.

Similarly, ChatGPT hit 100 million users in just two months after its late 2022 launch - a pace of adoption that we had not seen before with regards to any consumer products. The numbers reflect the preparedness and maturity of the masses for new product adoption with access to basic technologies and infrastructure already in place.

The Geopolitical Shift: From West to East

The transition from 2G to 3G to 4G wasn't just a technology story - it marked a significant geopolitical shift. Initially a Europe-US dominated landscape, the scene changed dramatically when leading Chinese players like Huawei and ZTE entered post-2010. They didn't just disrupt costs; they shifted the centre of gravity of telecom innovation eastward. European dominance gave way to Chinese leadership in 5G.

That pattern is eerily similar right now, but faster with Deepseek's arrival, barely two years after ChatGPT, along with new models from Alibaba and other Chinese tech giants, challenging the perceived US supremacy in AI. The arrival of these models has already reduced costs by a factor of 30 in just a short span. This isn't just about technology and it’s costs - it's about who shapes the future of global digital infrastructure.

Interestingly, India's role in this era was initially limited. While not a major infrastructure player early on, India emerged in the late 2010s as one of the most crucial consumer markets globally, driven by its massive scale of smartphone and internet users. This transformation of India from a peripheral player to a central market offers interesting possibilities for the AI era. With India's very different economic and technological positioning in 2025 compared to 2000 or 2010, could it play a more strategic role in balancing the US-China AI rivalry?

A crucial pattern seems to be repeating (yes pattern matching is always fun)

• The TelCos and their infrastructure eventually turned into a commodity despite laying out rails of the digital economy.

• Value pools were captured by the OTTs (Over The Top players) from social networks to content leaders like Netflix and Spotify under the hood of the App store ecosystems laid out by Appel’s iOS and Google’s Android.

• The combination of infrastructure rollout (2G to 3G to 4G), smartphones, and App store ecosystem put the world in full momentum for internetification from early 2010s.

• Now, in just two years, we're seeing the LLM infrastructure - the rails of the GenAI infrastructure - starting to stabilize and commoditize much faster than it happened then.

• The enterprise AI solution providers are already starting to provide solution that are LLM agnostic and offer interoperability (some remember the arrival of number portability and MVNOs from that telco era?).

• The consumer tech wave for new AI products for end users seems to be kickstarting where the value will be captured by this application layer and may not accrue at the infra (LLM) provider.

The Agentic Revolution: A Fundamental Shift

If you try to chart out the above pattern further. The most transformative aspect of the AI era will most likely be the emergence of autonomous agents - a paradigm shift that goes beyond anything we saw in the internet era. Unlike apps that wait for our commands, AI agents can proactively execute tasks, learn from interactions, and work together.

In the enterprise space, this means rethinking everything we know about software and services. The SaaS model that dominated the cloud era will undergo a fundamental transformation (more on this in another piece about SaaS 2.0). Instead of humans interfacing with software through predefined workflows, we're moving toward agents that understand intent and execute complex sequences of actions.

For consumers, just as we now have dozens of apps on our phones, we'll soon interact with multiple specialized agents handling different aspects of our lives - from managing schedules to executing transactions, from creative work to personal assistance. These agents won't just process information; they'll take actions on our behalf.

The Accelerated Timeline

The compression of timelines in the current wave is remarkable.

The Internet Era (2000 to 2020).

• Telco infrastructure build-out: ~8-10 years (early-mid 2000s)

• Smartphone revolution: ~6-8 years (2007-2014)

• App ecosystem maturity: ~6 years (2009-2015)

• Chinese infrastructure disruption: ~5 years (2010-2015)

The commoditization of core infrastructure and rails by the TelCos, from seeming overinvestment in infrastructure to losing value capture to OTTs, took over a decade.

The GenAI Era (2020/22-present):

• Pre-GenAI foundation building (2020-2022)

• ChatGPT launch and consumer GenAI arrival (Late 2022)

• Multi-player race emergence (OpenAI, Anthropic, Gemini, Llama)

• Early LLM commoditization with Deepseek and Chinese models (2024)

• Dramatic cost reduction (30x decrease in months)

What took 20+ years in the internetification of the world (1998-2020) might take just 8-10 years (2022-2030) in the AI era, driven by:

• Pre-existing digital infrastructure

• Mature global supply chains

• Faster technology adoption cycles

• Greater capital availability

• More sophisticated developer tools

The Enterprise Transformation Ahead

For enterprises, the implications are profound:

• How do you position for an AI-first world when the technology is evolving monthly?

• What happens when every business process has AI agents as active participants?

• How do you prepare for Chinese AI companies potentially disrupting current cost structures?

The most intriguing question might be about market leadership. While today's AI landscape seems dominated by a few major players, remember that the early internet era's leaders weren't necessarily those who ended up dominating the mobile internet age. The transformation of enterprise software through SaaS was a defining story of the internet era - what's the equivalent for the AI age? Who will dominate the Agent era? Consider Salesforce's Agentforce - could this be similar to how Microsoft transitioned from desktop Windows to cloud leadership after losing the mobile OS battle? Or is this just a hype?

The Consumer Layer: New Interfaces, New Experiences

For consumer products and applications, here's a crucial question: Just as smartphones revolutionized internet interaction, what comes next for AI? Could we see a shift from screen-based interfaces to ambient computing? Are Meta's Ray-Ban smart glasses pointing the way? A dialogue-based interactive UX is one possibility I find convincing.

The smartphone app ecosystem, dominated by Google's Android and Apple's iOS, transformed digital services. What's the equivalent for the AI era? Agents for personal use? What would be the form factor of these Agents? Who will be the Meta or TikTok of the AI age?

Critical Questions to ponder over as we navigate ahead

While the parallels with the internet era offer valuable strategic and tactical insights for many of us who've personally implemented and closely observed these generational transformations, the AI revolution brings its own unique challenges and uncertainties. Here are some questions we will need to consider and the challenges that we will need to overcome as we building our strategies for the next few years and then executing them despite all the uncertainties:

• Timeline and Adoption: Will AI evolution truly follow this compressed timeline? What unique technical, regulatory, or social barriers could emerge that we didn't face in the internet era? The governments and regulators are much more circumspect this time - what speed bumps would they put in place to ensure safe speed limits are adhered to?

• The Agentic Shift: How will the trust and adoption patterns for AI agents differ from app adoption? What new challenges exist in convincing both enterprises and consumers to delegate real actions to AI agents?

• Enterprise Integration: How will AI integration challenges differ from previous digital transformations? What new organizational and technical hurdles will appear beyond what we saw with SaaS adoption? The importance of data access , security and privacy to set the right guardrails is top of mind for organizations.

• Interface Evolution: Do we really need new interface paradigms for AI interaction? Can existing interfaces evolve rather than be replaced?

• Workforce Transformation: How can we better prepare for potentially more dramatic workforce changes than we saw in the internet era? What new skills and adaptations will be crucial?

• Business Model Innovation: What fundamentally new business models can be churned out? Outcome based? Service as a Software? Usage based? Monetization strategists are in a hyper drive mode right now to spin new and creative approaches. While many might be combinations of earlier models, how will AI ownership and deployment models differ from SaaS 1.0?

• Geopolitical Dynamics: How will AI development and adoption be shaped by current geopolitical tensions? Will we see a more fragmented AI landscape than we did with internet technologies?

• India's Evolving Role: Given India's massive tech talent pool and digital-first economy, could its role in AI development be fundamentally different from its internet era journey? Could India play a critical role with its very different economic and technological positioning in 2025 than what was in 2000s?

These questions highlight why this transition, while rhyming with history, will require fresh thinking and adaptability. If you have actively participated in the navigation of the internet transformation then perhaps you have an advantage to anticipate and address these challenges, but we must remain open to new paradigms and unexpected developments.

The Human Impact. Not just being in the loop.

This transformation isn't just faster - it's fundamentally different in how integrated the technology stack is and how quickly capabilities are evolving and the roles and responsibilities collapsing.

The next few years promise to be even more transformative than the internet revolution - and we're all part of this extraordinary journey.

The impact on workforce and careers will be profound (a topic I hope to explore in more detail in an upcoming post). What excites me most is the potential for entirely new interaction paradigms and business models. Just as few of us in 2007 could have predicted how profoundly smartphones would change our lives, we're likely standing on the edge of an equally revolutionary changes in how we interact with technology and each other.

Would love to hear your thoughts. What parallels and other patterns you see emerging and what big overlaps and differences I missed in this comparison?