History has a way of coming back to you now and then.
Take, for example, the empires of history; there are loads of them at this point. The Roman Empire. The British Empire. The Mongol Empire and so on!
Of these, the Roman Empire wasn’t the biggest or the longest-lived — but its impact? Immense and immovable.
We still live in its shadow: in our laws, philosophies, political systems, and art.
Which makes you wonder: why did Rome’s influence last so long?
When you think about it long enough, a central theme appears: Great empires are built upon enduring principles and a self-sustaining culture.
In the world of tech, no other company exemplifies following “enduring” principles better than Google!
Google’s principles, as I’ve understood them, are: focus on core fundamental scientific research, making this research universally accessible and useful, without being evil.
Just a year ago, when OpenAI was considered the frontrunner and everyone was poking fun at Google for being slow to innovate and constantly playing catch-up.
Now, fast-forward twelve months, and the story has shifted. The Gemini platform is now rock solid. Google’s enterprise-AI suite is among the strongest in existence.
With Alphabet’s deep pockets, its moon-shot projects, and a raft of yet-to-monetise products, the company could sail this wave for a very long time.
While every other company is building pretty castles with their pretty little moats, Google’s the one building an empire.
And in this newsletter, I’ll explain why I believe Google will win the AI race.
Btw, we’ve just started posting on YouTube! It’d mean a lot if you subscribed.
We’re releasing these videos alongside their written tutorials. Get actionable insight in the form of videos.
Please do subscribe!
Before we dive in, a small confession:
I’ve always had a soft spot for Google. Not just as a techie, but as someone from India. In developing countries, Google has done more to democratize knowledge and resources than almost any other company.
Unlike Meta or Microsoft, Google’s impact feels deeply infrastructural — like roads and libraries built for the digital age.I use “Google” and “Alphabet” interchangeably. In post cases, I do mean Alphabet, but you’ll understand.
To win the AI race, a company must dominate three pillars:
Compute
Talent
Data
Add to that the softer advantages — goodwill, distribution, and cash reserves — and you’ll see: Google has all of them, in excess.
1/ Compute Supremacy
While the rest of the world is paying a 60-times-earnings multiple to buy Nvidia stock because they’re the “only ones that have a production GPU that do ML OPs with CUDA,” it needs to look into Google.
While the AI cloud companies (as well as the country of China) are dying for the Nvidia GPUs, we have Google that has been quietly building its own compute empire.
Since 2015, they’ve been using their own in-house GPUs, specifically designed for ML-ops called TPUs, to work with their own data and ML workloads.
TPUs are contextually better than GPUs. Why?
TPUs are faster (1.2x - 1.7x), consume less power (1.3x - 1.9x) for AI operations than GPUs because TPUs are custom-built for only one purpose — to crunch AI workloads, while GPUs are general-purpose parallel processors used for minting crypto or playing games, or rendering 3D images.
Because Google owns its entire stack — from hardware design to TensorFlow integration and cloud orchestration, it allows tighter optimisation similar to Apple’s ecosystem approach. This results in better efficiency and scaling within Google Cloud environments.
Meanwhile, for Nvidia GPUs, AI companies often rely on additional optimisation frameworks, CUDA libraries, or third-party accelerators to fully leverage GPU performance.
This vertical build-up gives Google a strategic advantage in every AI compute bottleneck that constrains other AI players — design, deployment, power, and economics.
This is Google’s moat. Available to anyone rich enough to pay for an AI Cloud and Compute service provider!!!
And, just yesterday, Google made the latest Ironwood TPU available for public use via Google Cloud!!
2/ Data Supremacy
Data is the fuel of AI, and Google sits on one of the richest troves.
Just the breadth of data that it has access to — search queries, maps, videos, text, images, book scans, user behaviour, device signals.
But, much more than the amount of data, it’s the quality of data that matters.
Google has quality data in two ways:
Being a search engine first, all of Google’s data is annotated, well-structured, and linked.
This is the quality of data that companies like ScaleAI get acquired for.
All of Google’s data is real-time. Using the data on the internet to form an ever-evolving mind-map (called Knowledge Graph) is an entirely different use case that Google has mastered over the years.
In fact, every real-time or search query made on Perplexity or OpenAI, under the hood, uses Google search APIs — albeit with some added steps.
Now, there are some troves of data Google has that the rest of the internet doesn’t have access to:
Google Books: Google has digitised tens of millions of volumes of books, across languages and times, giving access to textual data far beyond what one finds on the open web. Anthropic had to scrape pirated books to get access to this data.
Active and passive user signals data: Data collected by Google’s services continuously over time, which is instrumental in understanding humans and how they work. (This is the AI race, all about, btw)
Multi-modal Connected Data: When a user is connected to YouTube, Android/Chrome, Maps, Gmail, and more, all at the same time, and you have a system that connects each data-point across these modalities (text, image, video, behaviour). This type of data is GOLD for the scale for AI.
The quality of data and the trove of data that nobody even has access to give Google the unique ability to build “real-world” generative AI systems grounded in rich data from the planet, not just the web.
3/ Talent
No company holds the prestige and goodwill in the tech community as Google does.
Under Google, DeepMind, a premier AI research firm they acquired in 2014, made AlphaGo an AI system that beat a human in Go, and paved the way for a new way of training machines, called reinforcement learning.
In 2020, post the onset of COVID, the same team solved the protein folding problem that could help researchers create a vaccine for the next pandemic faster. For this, they were even awarded the famed Nobel Prize for Chemistry.
They were the creators of the transformer architecture in 2017, which is the backbone of all AI LLMs today.
Beyond building models, they publish open-source tools (TensorFlow, BERT) and deploy research into production systems.
This research-intensive culture means Google isn’t just building AI features — it’s expanding what is possible in AI.
That gives it a structural advantage in attracting and retaining elite talent, solving hard foundational problems, and moving from research to deployment faster than many peers.
For any one company to be the patron of these AI researchers on you have to have deep conviction and focus seeping from its leadership.
Google has both.
For over a decade, Google has been focused on AI and AI only. There are instances of their AI focus from 2016 and 2017.
And with Sundar and Demi Hassabis at the helm, you couldn’t ask for better leadership.
4. Miscellaneous: Current Money and Future Potential
Google already makes a lot of money. It made ~102B in the last quarter only. First company ever to do $100B revenue in a quarter.
But, if you look closely, there’s only a handful of its existing projects that this revenue is coming from.
This is what is mean when I say it’s an Empire.
It hasn’t even opened revenue taps on so many of its projects! It can never run out of products to monetise; they’ve ensured it.
A. Cash On Hand
Google currently holds $95.14 billion USD in cash on hand.
Now, this figure does not represent the company’s full asset base — its data centres, power infrastructure, investments in long-term projects, but if Google needed to raise new capital, for any of its AI subsidiaries, it could be viewed as having more credibility than many smaller challengers.
Considering the hyper-valuations of AI companies, Google does seem like an AI hedge against the AI bubble speculations.
B. Unexplored Revenue Streams
Google has to be one of the very singular companies that has many potential revenue streams that haven’t yet been fully unlocked.
In the age where AI is making the entire value prop of many tech companies moot — hello Adobe — we have an endless stream of successful projects with Google that have yet to be monetized.
Think Google Maps, that’s still free right now.
Think Google Workspace tools and APIs that are still under-explored and have generous free tiers.
Then there are tools like AdSense, Google Analytics, Google Sites, etc, that are still available for free.
These latent assets give Google the option — at the right moment — to open up entirely new revenue lines, placing the company in a very strong strategic position.
C. X/Moonshot Projects
Google’s Moonshots are high-risk bets that Google takes on speculative ideas.
Think Waymo, a self-driving robo-taxi startup in a $300B mobility market space. It began as a Google Moonshot project in 2009, and after 16 years of research, live in 5 US cities!
Similarly, there are projects like Wing for autonomous drone deliveries.
But these are graduated ideas.
There are then ideas that they’re still working on:
Project Suncatcher: A space-based, AI infra with satellites equipped with TPU AI chips. These satellites would harness near-continuous solar energy, potentially making AI data centers in orbit many times more power-efficient than Earth-based centers.
Google’s Quantum AI Bet: Like with AI, Google is years ahead of its peers in quantum computing. Leveraging its deep experience in TPU chip design, the company launched the Willow quantum chip — and recently demonstrated verifiable quantum advantage — proof that its quantum processors can outperform classical supercomputers on real workloads, marking a major milestone in the field.
And many others are listed on their official page, ranging from biology to material sciences to robotics.
Google is indeed a technical empire.
BONUS: Some Grain of Salt
All said and done, do take all of this with a decent grain of salt. Google has historically been very ruthless with its products. They’ve killed great ideas either due to bad execution or due to challenges post experimentation.
Google’s innovation culture has a flip side, too.
A long list of moonshots that never made it to market. From Google Glass to Loon, the company has sunsetted dozens of projects due to poor market timing or lack of adoption.
For developers, that unpredictability is always a risk when building within Google’s ecosystem.Execution Woes
Google has often struggled to turn bold ideas into cohesive products. Its messaging history is a case study in fragmentation — from Allo, Duo, and Hangouts to the eventual consolidation under Gemini.
Google+, many video calling tools during COVID, Stadia, the list goes on!
Google’s scale and experimental culture often lead it to chase parallel ideas without clear integration, diluting focus. Yet, paradoxically, this same willingness to fail fast is what keeps the company at the frontier of innovation.
Closing Thoughts
Google remains the company most others aspire to become.
When the AI gold rush cools, the winners will be those who built foundations, not fads. Google’s vast data, compute stack, and research culture give it a structural edge few can match. It’s less a company competing in the race, and more the ground on which the race is being run.
That’s the kind of empire worth betting on.
And with the tech market recalibrating in fears of a bubble forming, Google stands out as one of the more legible, durable bets in the market.
Not financial advice — just where I’m placing my chips.
Anyway, that’s it for this week’s newsletter.
Until next time.
I have to admit I was one of those poking fun at Google for being slow to innovate and constantly playing catch-up. I even wrote about it here. Of late I am beginning to realise just the sheer volume of Google tools available and the volume of public data Google has in its possession, Google has a clear chance to lead trends...