Stop Burning Cash on “Smart” AI: Why Your Business Needs a Dumb Delivery Van
This article was based on an interview with Mr Vincent Caldeira, Chief Technology Officer at Red Hat and Ms Tammy Tan, Country Manager at Red Hat, Malaysia.
We have been sold a lie.
For the past year, the tech industry has been gaslighting businesses into believing that “bigger” equals “better.” The narrative is seductive: if you aren’t running a Foundation Model trained on the entire internet with trillions of parameters, you aren’t really doing AI. You are falling behind. You are irrelevant.
Image Generated by Gemini
This is complete nonsense. And it is a financial disaster waiting to happen.
If you are a local retailer in Malaysia, you do not need a model capable of writing a symphony or coding a video game to tell a customer where their parcel is. Buying a trillion-parameter model for customer service is the business equivalent of buying a Ferrari to deliver a pizza. It looks incredible, it sounds expensive, and it destroys your profit margins with every mile.
The future of AI isn’t the supercar. It is the beat-up, reliable delivery van.
To understand why the smart money is moving away from the “wow factor” and toward “industrial efficacy”, I spoke with Tammy Tan, Country Manager of Red Hat Malaysia, and Vincent Caldeira, Red Hat’s Chief Technology Officer for APAC. They laid out a brutal reality check for Malaysian businesses: stop chasing the hype and start looking at the unit economics.
The Trillion-Dollar Hole
Let’s look at the money, because the math isn’t adding up. The industry has dug a “trillion-dollar hole” in global AI revenue. We see massive capital expenditure on data centres and training, but the revenue return is nowhere near covering it.
Vincent Caldeira points out that the market is currently rigged by a few giants.
Source: Red Hat
“Currently, the situation is that closed models (OpenAI, Anthropic, Google) capture 96% of revenue and 80% of usage of all AI inference revenue… creating a market dominated by a few major players with significant pricing power.”
Vincent Caldeira, Chief Technology Officer (APAC), Red Hat
This is a dangerous trap for Malaysian SMEs. When a handful of US tech giants control 96% of the revenue, they control the price. If your business relies on renting intelligence from a closed API, you are exposed to “aggressive API price hikes” and “technological dependency”. You are building your house on someone else’s land.
The Case for the “Dumb” Van
The solution is to stop renting the Ferrari and buy the van. This means adopting an open-source strategy with smaller, specialised models.
The financial difference is staggering.
Source: Red Hat
“If we only focus on inference cost at the model level, a recent study found that open models are approximately six times cheaper on a usage-weighted basis than closed models.”
Vincent Caldeira, Chief Technology Officer (APAC), Red Hat
Six times cheaper. That is the difference between a project that bleeds cash and one that scales sustainably.
But you don’t just pick a small model and hope for the best. You need “system-level efficiency”. Caldeira describes a strategy called “Intelligent Model Routing”. Think of it as a traffic cop inside your software.
If a customer asks, “What time do you open?”, the traffic cop sends that query to a tiny, cheap model. If a customer asks, “Analyse this legal contract,” it routes it to the expensive “Ferrari” model.
Source: Red Hat
“…a system can automatically direct simple queries to lightweight expert models while reserving expensive high-reasoning models only for complex tasks, ensuring you never overpay for unneeded intelligence.”
Vincent Caldeira, Chief Technology Officer (APAC), Red Hat.
Escaping “Pilot Purgatory”
Even if you pick the right model, you will likely fail if you treat AI like a science experiment. Tammy Tan highlights a depressing stat: while half of SMEs expect AI to drive growth, only 1 in 5 has successfully integrated it.
Most companies are stuck in “Pilot Purgatory”. They have a cool demo running on a laptop, but they can’t get it to work in the real world.
Source: Red Hat
“A single server PoC doesn’t reflect the complexity of production-grade AI. Once you scale, you deal with unpredictable inference workloads, different hardware profiles, data privacy requirements, and strict service-level objectives.”
Tammy Tan, Country Manager, Red Hat Malaysia
You cannot solve this by “just adding more machines”. You need an “intelligent infrastructure layer”—an AI-aware control plane that manages your compute and memory like a factory floor, not a laboratory.
The Energy Bill You Can’t Ignore
There is another reason the “Ferrari” strategy fails: electricity.
We often ignore sustainability as a PR metric, but in AI, energy is a direct cost. Caldeira drops a critical fact: 90% of energy is spent on inference, not training.
The day-to-day running of your chatbot costs vastly more than teaching it. If you are running a massive model for every interaction, you are burning watts—and ringgit—unnecessarily.
Source: Red Hat
“Most Malaysian companies… are becoming more aware that AI has a real energy cost behind it, not just from a sustainability angle, but from an operational budget perspective.”
Tammy Tan, Country Manager, Red Hat Malaysia
By using optimisation techniques like quantisation (making the model smaller), you can run these systems on standard commodity hardware instead of hunting for scarce, expensive GPUs.
Source: Red Hat
“Platform-level optimizations like caching and quantization can achieve up to an 800x reduction in energy footprint.”
Vincent Caldeira, Chief Technology Officer (APAC), Red Hat
Boring is Profitable
We need to grow up. The “wow” phase of AI is over.
Tan predicts that by 2026, the market will shift entirely to “industrial efficacy”. The winners won’t be the companies with the coolest demos. The winners will be the ones who made AI boring, reliable, and cheap.
She points to the financial services sector as the likely leader. Banks deal with high-volume, rules-driven processes. They don’t need creativity; they need accuracy and auditability.
Source: Red Hat
“It may not make headlines, but it will deliver real business transformation the soonest.”
Tammy Tan, Country Manager, Red Hat Malaysia
Kill the Ferrari, Save the Business
The party is over. The “wow factor” that sold us on AI last year is now the very thing threatening to sink your ROI.
We are seeing a brutal pivot from experimentation to survival. The “trillion-dollar hole” isn’t just an abstract statistic; it is a graveyard for companies that treated AI as a toy rather than an industrial tool.
If you are a CEO or IT leader in Malaysia, you need to stop chasing benchmarks. The “smartest” model on the planet is useless if it bankrupts you. You need the model that is smart enough, the architecture that insulates you from price gouging, and the sovereignty to own your data.
Source: Red Hat
“Business value only emerges when the model is embedded in a complete, secure and scalable AI system.”
Tammy Tan, Country Manager, Red Hat Malaysia
Don’t buy the Ferrari. Buy the fleet of reliable, efficient vans. Your shareholders—and your energy bill—will thank you.
This article was based on an interview with Mr Vincent Caldeira, Chief Technology Officer at Red Hat andMs Tammy Tan, Country Manager at Red Hat, Malaysia.
Vincent Caldeira Chief Technology Officer, Red Hat (APAC)
Vincent Caldeira is Field CTO, APAC at Red Hat. In this role, he is primarily responsible for engaging and building partnerships with strategic customers to explain Red Hat’s vision and establish and reinforce Red Hat as an industry leader while establishing trusted relationships with customers’ technology leaders and advocating for relevant emerging technologies.
Vincent has spent more than 20 years of his career in the financial technology sector, both as a Chief Technology Officer shaping technology strategy, enterprise architecture and driving technology transformation roadmaps, as well as driving talented engineering teams to design, build and deliver software solutions in the financial software vendor industry.
Vincent also contributes to OS-Climate, a Linux Foundation-backed open source project that intends to build the breakthrough technology and data platforms needed to more fully integrate the impacts of climate change in global financial decision-making and risk management, where he acts as the lead architect and Technical Advisory Council member.
Vincent holds a Master of Science in Management (majoring in Information Systems Management) from HEC Paris.
Tammy Tan Country Manager, Red Hat Malaysia
As country manager, Tammy Tan is responsible for overall sales and business aspects of Red Hat Malaysia. In this role, she leads the country team to drive sustainable growth by building strong and collaborative relationships with enterprise customers, as well as partners.
Tammy joined Red Hat from Cisco Malaysia, where she served as the Director of Enterprise and Commercial. She brings with her 21 years of experience in the IT industry, with deep experience in Financial Services, as well as a broad perspective on vertical solutions within the Energy, Oil and Gas Sector, Transportation, Manufacturing and Hospitality. Tammy has a passion for STEM and started her career as one of the first female engineers at Siemens and NTT Communications. She holds a degree in Electrical Engineering from the Multimedia University of Malaysia.
An avid tech enthusiast and passionate about what comes next. Jeevan has been entwined in tech since he was in high school and has a background in Medical Bioscience. He's passionate about tech and more importantly what benefits can come from it.
