AGI On The Horizon: Will Your Business Be Ready Before It’s Too Late?
- Zac Engler
- Feb 24
- 3 min read
In 2016 I estimated that AI would reach human-level capacity in 2027. In 2019, I re-estimated that AI would reach 4.33 exaFLOP/s—four times the estimated compute of the human brain—by 2028. That projection felt aggressive at the time. Today, it might be too conservative.
El Capitan: Supercomputers Have Surpassed the Human Brain
In November 2024, the El Capitan supercomputer claimed the #1 spot on the TOP500 list, delivering 1.742 exaFLOP/s—exceeding human brain-level compute. This milestone, once thought to be a decade away, has already arrived.
However, raw compute alone does not create AGI; advanced AI models do, and that is where Grok 3 comes in.
Grok 3: Smarter Reasoning, Faster Development
The former tweeting company, xAI launched Grok 3, a state-of-the-art LLM trained on Colossus, a supercomputer using over 100,000 NVIDIA GPUs. Andrej Karpathy, a former OpenAI executive, tested Grok 3 and found:
Its “Thinking” mode rivals OpenAI’s top-tier models, outperforming DeepSeek-R1 and Gemini 2.0 Flash Thinking on logic-based tasks
It solved complex reasoning problems that many AI models struggle with
Unlike most AI models, it attempted complex mathematical proofs rather than defaulting to “this is an unsolved problem”
Most importantly, xAI built Grok 3 in under a year. This means that AI models are now improving at the same pace—or faster—than the hardware they run on.
The Competitive Landscape: OpenAI and DeepSeek’s Rapid Advances
The AI race is intensifying, with major players accelerating their developments:
OpenAI is set to release GPT-4.5 within weeks and GPT-5 within months, as announced by CEO Sam Altman. These iterations promise enhanced capabilities, pushing the boundaries of AI performance. 
DeepSeek, a Chinese startup, has introduced DeepSeek-R1, an AI model that matches top-tier performance while being more energy-efficient and cost-effective. This model has been rapidly adopted across various industries, including finance, where firms like Tiger Brokers have integrated it into their operations.
AGI Timelines: The Future Just Got Closer
If AI progress continues at this speed, we could see AGI as soon as 2026-2027, rather than the previously expected 2028-2030 window. The combination of:
Supercomputers surpassing human brain-level compute
LLMs improving reasoning, adaptability, and self-correction
The fastest iteration cycles in AI history
…suggests we are closer to AGI than nearly anyone expected just five years ago.
The Business Reality: Adapt Now or Get Left Behind
AI is not just about technology; it is about economics, competition, and survival. The global AI market is projected to reach $826.7 billion by 2030, but unlocking real economic growth will require $1 trillion in AI investment by 2025. The companies that act now will dominate the next decade.
To stay ahead, executives must:
Implement AI today; companies that wait will lose market share
Leverage AI for strategic decision-making; it is no longer just about automation
Consider self-hosted or Soverign AI solutions; ensure global shifts don't disrupt you
Trifurcate your workforce strategy; guide teams to collaborate AI not fear replacement
TLDR: AI is Moving Faster. Your Business Needs to Keep Up!
We are beyond speculation. The infrastructure, compute power, and models for AGI are being built right now. Companies that treat AI as a future concern instead of a present priority will not catch up later.
At Bodhi AI, we help businesses integrate AI workflows, optimize decision-making, and stay ahead of industry shifts. If your AI strategy is not in motion today, you are not just behind; you are compounding risk while others compound advantage.
Comments