As the model race becomes a contest for scarce resources, the AI industry is being led down an unsustainable path. Today, a handful of companies control the chips, capital and data centers required to build today’s largest AI models. The likes of Alphabet, Amazon, Meta and Microsoft are on track to spend nearly $700 billion on AI infrastructure this year alone. This is more than triple what they spent just two years ago.
A single Nvidia GPU can cost up to $40,000. Extrapolate that out to the hundreds of thousands needed to train frontier models, and the total runs into the billions. With data centers consuming as much electricity as a mid-size city, what was once a software industry is starting to look like utility-scale infrastructure. Unfortunately, t
hese staggering capital commitments are not yet translating to success. In fact, one recent MIT report found that a staggering 95% of corporate generative AI pilots fail to deliver measurable financial returns. Something needs to change.
AGI 2.0 and the shift to smarter training
Artificial general intelligence (AGI) 2.0 represents a turning point in how intelligence is built and, critically, where the next investment frontier lies. While the tech giants double down on capital-intensive scale, a different trend is gathering momentum. Low-cost open-source models are accelerating. China is setting the pace with Alibaba’s Qwen leading global downloads, user preference and new model adoption. In September 2025, almost two-thirds (63%) of all new model adoption came from Chinese systems, more than double the US share. This is prompting investors to question whether relentless frontier‑scale spending is still the way to go.
The scientific community is reaching a similar conclusion. David Silver, former DeepMind researcher, launched Ineffable Intelligence in late 2025 on the thesis that large language models (LLMs) trained on human data represent a ceiling. His company raised an impressive $1.1 billion in April 2026 at a substantial $5.1 billion valuation. Turing Award winner Yann leCun has made an equivalent wager. Having publicly argued for years that autoregressive LLMs are a dead end, LeCun left Meta last year to found AMI Labs, raising $1 billion at $3.5 billion valuation in March 2026.
Investors are paying attention because the economics are changing. Training costs are a major factor. Since 2016, they have risen roughly 2.4× a year, from tens of thousands to tens of millions of dollars and are projected to exceed $1 billion by 2027. Frontier-scale models are increasingly constrained by their own cost curves.
Smart data, not big data
There is no doubt that markets are moving away from “train-everything” general-purpose systems toward domain-adapted models. The evidence is clear. Smart data beats big data every time. In early 2025, Li Fei-Fei’s team at Stanford fined-tuned Alibaba’s Qwen 2.5-32B-Instruct model on a curated dataset at a cost of under $50, achieving performance comparable to top-tier reasoning models like OpenAI’s o1 in mathematics and coding capabilities.
AGI 2.0 goes even further. Models learn instantly, continuously adapting rather than being retrained from scratch. This reduces costs, speeds up deployment cycles and opens the door to intelligence that scales with use, not with spend.
The new economics of adaptive AI
When billion-dollar budgets become the barrier to innovation, opportunity collapses. Live learning changes the economics as it promotes sustainable intelligence creation. Experiments with the Boltzmann Learning Machine show that models learning as they are used can cut compute and energy costs by orders of magnitude. Matching or surpassing fine-tuned models with up to 22-point accuracy gains for only 10% more runtime and 2GB of additional memory. This means fewer retraining cycles, faster iterations and far more capital-efficient intelligence.
It is important to remember that AI’s most valuable asset isn’t compute, it’s the intelligence created through customer data, workflows and proprietary knowledge. Under AGI 2.0, that intelligence stays with the user. Models learn inside a company’s boundary and deploy in isolated environments, so private intelligence never becomes free fuel for another platform. In a world where increasingly stringent data-residency rules are, quite rightly, constraininghow and where training can happen, sovereignty becomes not just a principle but a competitive advantage.
Today’s AI stack is powerful but rigid. Layers of tools that don’t learn from each other. AGI 2.0, however, calls for systems that adapt not just to people but to other models, tools and environments. Interfaces reshape themselves through natural language. Knowledge flows across agents. Live learning replaces retraining. When intelligence adapts in real time, it becomes fluid, interoperable and self-improving.
Why scale still matters
Scale still wins benchmark tests and pretraining breadth but the economics are certainly tilting. Research has shown that each new generation of frontier models costs exponentially more than the one before while delivering smaller functional gains. Meanwhile, BOLT-style experiments demonstrate that live learning at inference delivers higher accuracy with minimal extra compute.
Adaptation flips the cost curve. Capability improves with use, not capital. Meta’s stock fell sharply after its Q1 2026 earnings report as investors focused on the scale of its AI spending plans which underlined the fragility of a strategy built purely on capex.
The next billion-dollar AI category
The biggest risk in AI is not missing the next model. It is losing ownership of the intelligence investors are paying to create. They are increasingly demanding clarity on three metrics. The share of spend going on retraining, the exposure to third-party data licensing, and the speed with which deployed models adapt. The companies that control their data, keep lineage traceable and adapt models “in flight” will attract the most investment as they will boastfaster cycles, lower unit costs and defensible moats.
As infrastructure costs soar and enterprise pilots struggle to deliver ROI, adaptive models offer a more profitable path beyond brute-force scale. The next winners in AI won’t be those who spend the most on compute but those who turn adaptation itself into the engine of scale.
