The Apostle Manthus
AI has already made fortunes on Wall Street. But if Goldman Sachs' predictions of 2030 are close to reality, what we have seen can only be the beginning.
We crawled on the huge wave of money and We ran into three Greek listed. Names matter, but the reason we found them there is even bigger.
In fact, if the tsunami arrives with the force the analysts expect, He'll literally take them and pick them up.
The Goldman Sachs, in its analysis "Decoding the Ancient Economy – The Coming Inflation in AI Usage and Marchins", estimates that we may still be together In infancy.
The use of AI tokens estimated to be launched up to 24 times by 2030, touching them 120 quadrillion tokens every month.
Before we proceed, we need a small translation, because these numbers usually address computer engineers rather than investors.
One token is a small information unit that reads or produces an artificial intelligence system. It can be a small word, a larger word section, a number or a punctuation point. Whenever we ask a model to read text, analyze economic data, write code or answer a question, they are consumed tokens.
For an investor there is an easier way to see it.
Tokens can be treated roughly like the work units passing through the AI factories. The greater the computational burden that needs to be served.
Quadrillion is even more foreign as a term. A quadrillion equals 1.000.000.000.000.000, a number with 15 zeros.
So what? 120 quadrillion tokens per month provided by Goldman means 120,000,000,000,000,000 units of processed information each month. Over 46 billion tokens every secondDay and night.
Goldman starts around 5 quadrillion tokens per month in 2026 and reaches 120 quadrillion in 2030.
Twenty-four times in four years.
Most people have met AI through a chatbot. They write a question, the model answers and the process ends.
The next cycle concerns AI agents.
An agent can take an order and perform it through dozens or hundreds of successive actions. Search for information, read files, open databases, compare results, write code, check his work, run a calculation again and use other programs.
The man asks a question. Agent can perform in background dozens of computing tasks to complete the command.
Goldman estimates that such a workload may consume 10, 20 or even 50 times more tokens from an ordinary chatbot conversation.
Think of a business with 50,000 workers. In a few years the same people may also use agents for accounting, programming, research, legal controls, sales, procurement, logistics, cybersecurity and customer service.
Each worker actually acquires one small digital staff working on servers.
And the digital staff doesn't sleep.
The detail of the greatest stock market significance lies in Goldman's assumptions. The provision of 120 quadrillion tokens in 2030 does not require universal adoption of agents.
Goldman estimates that then just about 12% of knowledge workers will have adopted agent AI.
Twelve percent.
24 times greater use with penetration of just 12%.
That's probably the number we need to get into our minds.
The following question now concerns both Wall Street and the Greek board.
What happens if agents become as common in business as in a decade of smartphones to consumers?
Attention, the usage data is already running. According to OpenRouter, the weekly volume increased from approximately 4.6 trillion in 113 trillion tokens within a year, nearly 25 times over, while only in the last month about doubled. More tokens ask for more GPUs, memories, networking, cooling, data centers and electric power. Goldman's 24x for 2030 already meets a real usage curve that rises at a dizzying speed.
AI looks immaterial on screen. But behind it is one of the largest industrial expansions in the last decades with huge demands on electricity.
And the first Greek stock comes right from there. The PPC.
Group plans in Western Macedonia data center of original power 300 MW, capable of expansion up to 1 GW, if agreements are concluded with hyperscalers. At its full size it will be among the largest data centers in Europe.
Now put two numbers next door.
24 times higher consumption AI tokens by 2030 according to Goldman Sachs. Up to 1 GW data center from PPC.
1 GW, if operated theoretically in full load throughout the year, corresponds to 8,76 TWh electricity. Size that explains why the data center race evolves simultaneously into an electric power securing race.
PPC has a remarkable package for Greek data. Production of electricity, RES, networks and project for data-center high power infrastructure.
The market is currently examining it mainly through electricity, RES, networks, Romania and dividends. The explosion of AI compute adds a new potential consumer of huge quantities of energy.
If Goldman approaches 24x, electricity becomes one of the most valuable raw materials in the new digital economy and PPC is already plugged in.
The Profile Software She plays on the other side of the same market and her case has become much stronger within 2026.
In 7 July 2026 presented the ProfileOne, a comprehensive platform organization for AI agent specially designed for financial institutions. The Timing perfectly fits with what Goldman describes about the next explosion of the agents.
ProfileOne allows creation and management of AI agents that can be connected with Profile systems, banking applications, CRM, data lakes, accounting systems and third party applications.
Profile attempts to put AI agents into the daily operation of banks, asset managers and financial institutions.
An agent can read data, communicate with different systems and perform sequences of tasks. This is exactly the type of application that can consume multiple tokens compared to today's chatbots.
Profile is active in more than 50 countries. If the AI is extended to financial services, the company already has products, customer base and international network on which it can add new technology.
For the shareholder, the question becomes very specific.
Can AI agents increase revenue per Profile client?
New licenses, subscriptions, recuring revenues, cross-selling and additional services on customers already using its systems can affect explosives the finances of a company this size.
Here is the big difference of 2026. ProfileOne transfers Profile from AI's general promise to a product that can be sold. And in a branch where compliance, risk management, portfolio management, reporting,lending and back-office functions offer huge field for agents.
For the PPC AI can mean more MW and data centers.
For the Profile more digital workers within financial systems and possibly more revenue per customer.
In Enterprise AI enters the game and the Quest Holdings (CUES) via Uni Systems.
The company created the Synthesis.ai, an AI Center of Excellence with applications already in more than eight sectors of the economy.
A bank, an insurance company, an industry or a large commercial enterprise needs cloud, corporate data, cybersecurity, integration and connection of agents to its information systems.
If millions of businesses move towards the agents in the coming years, the incorporation of AI into corporate information systems can develop into a market of huge size.
The money route finally brought us to three Greek listed in a completely different way of participating in the AI boom.
PPC can sell the energy and host part of the computer.
Profile can sell financial AI agents.
Quest can sell the corporate transition to AI and cloud.
And above that lies the prediction that began the entire debate.
120 quadrillion tokens a month. 24 times greater use. Only 12% penetration of agents into knowledge workers in 2030.
If 24x can occur with adoption just 12%, then in 2030 it may still be quite far from the full development of the agentic economy.
AI's first stock wave garnered enormous wealth around chips manufacturers and American technological giants.
The next will spread far further.
From token to data center. From data center to MW. From MW to network. From agent to banking software and from there to information systems of thousands of businesses.
Share charts you won't find today.
They'd make it all go away.
When Goldman talks about 24x by 2030The point is not the close resistance to the dashboard, But the size of the market coming.
* Apostolos Manthos is responsible for technical analysis & investment strategy
** The content of the Article may in no way be regarded as advice or suggestion or recommendation or invitation to purchase or sell any share or investment or financial product traded on an organised or non-market.
