An internal letter from the founder of Knowledge Atlas (ASX: 02513) details the company's strategic focus for the coming years.
The founder, Tang Jie, stated that the essence of the AI transformation is a leap in the "upper bound of intelligence." Artificial intelligence is progressing from perception and cognition towards Artificial General Intelligence, with the hallmark being the ability to create original knowledge on the level of "the theory of relativity." The trend from AGI to Artificial Superintelligence is irreversible and will reshape the entire technology stack and all industries. While the industry accelerates monetization, Knowledge Atlas is choosing to counter-intuitively "break upward," strategically investing in four core engines over the next two years: Long Horizon Tasks, Autonomous Agent Systems, Fully Self Training, and Extreme Safety and Governance.
On Friday, July 10th, shares of Knowledge Atlas closed down 19.29% at HKD 1,640, giving it a market capitalization of HKD 731.2 billion. This contrasts sharply with its position on June 22nd, when its share price reached HKD 2,410, rising 15.09% that day to achieve a total market cap of HKD 1.07 trillion, officially making it the first domestic large model company to enter the "trillion-HKD club."
From its HKD 116.2 IPO price on January 8th to exceeding HKD 2,400 per share and a market cap over HKD 1 trillion on June 22nd, and then retreating to HKD 1,640 on July 10th, Knowledge Atlas's stock price has traced a nearly parabolic trajectory in just half a year. This extreme volatility is not simply about gains and losses but represents an intense market game shifting from fervent consensus to reassessment of the AGI narrative.
The letter, titled "The Giant Wave Has Arrived," was addressed to the company's employees and partners concerned with the future of AI. It covered three main areas: the company's identity, its view of the current era, and its strategic direction.
Defining the Company's Core
The letter describes the company as one that does not chase trends. It grew from a laboratory, carrying with it a 20-year methodology summarized by three principles: essence, counter-intuition, and focus. It argues that deep thinking enables bold, unconventional choices, which in turn require long-term commitment. The company points to its history of seemingly "counter-intuitive" choices, such as early work on academic search systems and its significant bet on large-scale models well before the ChatGPT era. The core competency, it states, stems from a team willing to question fundamentals, dare to be counter-intuitive, and focus relentlessly on execution.
Assessing the Current Technological Era
The letter posits that genuine commercial opportunities lie not in incremental product or model tweaks but in the elevation of the "upper bound of intelligence." This current revolution is not merely product or business model innovation but a fundamental technological shift raising that upper limit. The path involves moving from perceptual to cognitive intelligence, and then towards AGI. The company defines AGI not as the intelligence of a single genius but as the sum total of human wisdom, capable of creating "theory of relativity"-level original knowledge.
The journey to this peak involves overcoming three major challenges: Long Horizon Task capability, Autonomous Agent Systems, and Self-Evolving systems. Mastering these would lead AI towards self-awareness, emotion, and potentially consciousness. The transition from AGI to ASI is seen as inevitable, driven by both algorithmic breakthroughs and the aggregation of massive computing power, and will reshape all layers of the technology stack and every industry.
Outlining the Strategic Focus
In light of this trend, the company's choice is again "counter-intuitive." While the industry broadly accelerates commercialization, Knowledge Atlas has decided to push upward. It has named this strategy the "Touch High Plan," aiming to challenge the current physical and algorithmic limits of technology over the next two years, with strategic investment focused not on short-term application monetization but on the next high ground of AGI.
This investment will concentrate on four core engines: advancing AI from instant Q&A to managing grand projects; evolving from intelligent assistants to digital employees via autonomous agent societies; pursuing Fully Self Training to use computing power as fuel for evolution as human-quality data becomes scarce; and implementing Extreme Safety and Governance, embedding ethical and legal norms into the model's core functions and investing heavily in interpretability to ensure the technology's safe and beneficial development.
Commitment to an Open Ecosystem
The letter emphasizes that the long-term development of AI as a strategic technology depends on an open, collaborative industrial ecosystem. True safety, it argues, comes not from technological secrecy and barriers but from widespread co-construction, sharing, and supervision. Reflecting this belief in technological accessibility, the company recently released its most powerful open-source model to date, GLM-5.2, with a permissive MIT license, making advanced capabilities available to all developers. This open approach is presented as complementary to the "Touch High" ambition—one hand reaching for the limits of intelligence, the other paving the way for its broad and equitable distribution.
Final Thoughts and Motivation
The letter concludes by addressing why the company, post-IPO, continues to pour core resources into this highly uncertain "Touch High" direction. The answer lies in a belief that true pioneers build roads up the mountain they climb. The company aims to build this road higher—high enough to ensure safety and explore the unknown—and wider, so that every developer and team can walk upon it. In the AGI era, these once-distant goals are now within reach, representing both great fortune and profound responsibility for this generation. The giant wave has arrived, and Knowledge Atlas intends to be the one riding it upward.