AI: The End of SaaS & Rise of Research Accelerators
An analysis of AI's transformative impact on business, investing, and the future of work, predicting the demise of traditional SaaS and the automation of knowledge.
Key Insights
-
Insight
The era of data labeling companies is over; it's now the era of 'research accelerators' focused on training superintelligence.
Impact
This signals a market shift towards more sophisticated AI development, requiring partners with deep research DNA and the ability to generate complex, real-world data, moving beyond basic labeling services.
-
Insight
Traditional SaaS, as we know it, is over, due to AI making custom software development easier and foundation models moving into the application layer.
Impact
Investors should reconsider long-term SaaS valuations, as companies may increasingly build in-house AI-powered solutions or rely on agentic models, potentially disrupting established software markets.
-
Insight
AI's progression involves a fundamental shift from teaching models to pass tests to enabling them to perform complex, economically valuable real-world work through agentic systems.
Impact
This paradigm shift drives demand for advanced data generation, particularly reinforcement learning environments, creating new opportunities for specialized AI data and deployment companies.
-
Insight
Enterprise AI adoption is hindered by 'first mile schlep' challenges, including fragmented data, lack of infrastructure for evaluations, and the need for new human-AI collaborative workflows.
Impact
Companies that can provide comprehensive solutions addressing data structuring, custom model fine-tuning, and 'hand-holding' through deployment will capture significant market share in enterprise AI.
-
Insight
The slow pace of AI adoption by incumbent companies will lead to a significant 'transfer of value' to agile, AI-native startups.
Impact
This suggests a period of market disruption where established players risk being outcompeted, creating prime investment opportunities in innovative startups leveraging AI for core business functions.
-
Insight
The immense 'grand prize' of achieving Artificial General Intelligence (AGI) compels major tech companies to make massive investments, viewing the cost of not winning as prohibitive.
Impact
This fuels hyper-competition and large capital allocation in frontier AI development, ensuring continued rapid advancement and high-stakes market dynamics among the biggest players.
-
Insight
AI development will experience a 'slow and steady takeoff' rather than a rapid, disruptive AGI arrival, providing humanity time to adapt.
Impact
This perspective suggests that value will be realized incrementally, potentially mitigating immediate massive job displacement and allowing for strategic planning in education and workforce reskilling.
-
Insight
The future of human-computer interaction will move beyond phones to multimodal, ambient AI devices (wearables, smart implants) acting as extensions of the human brain.
Impact
This opens new markets for hardware and software innovation in wearable technology, sensor development, and AI interfaces that continuously process and respond to multimodal human input.
Key Quotes
"I think the era of data labeling companies is over. And it's now the era of research accelerators."
"SAS as we know it, I think it's over. I think it's completely over."
"I don't see an AI bubble. These models are incredibly powerful today."
Summary
The AI Tsunami: Reshaping Business, Investing, and the Future of Work
The landscape of technology and business is undergoing an unprecedented transformation, driven by the relentless progress of Artificial Intelligence. Gone are the days of simple data labeling; we are entering the era of 'research accelerators,' where the objective is nothing less than training superintelligence. This shift brings with it profound implications for how companies operate, what investors prioritize, and the very nature of human work.
The Demise of Traditional SaaS and the Rise of Agents
Jonathan Siddarth, CEO of Turing, articulates a provocative vision: "SAS as we know it, I think it's over." He posits that the burgeoning power of AI models, especially agentic systems capable of executing complex multi-step workflows, makes it increasingly easy for companies to build custom software. This capability, combined with the foundational models moving into the application layer, threatens the traditional GUI-driven SaaS paradigm. Instead of relying on numerous third-party tools, businesses will leverage AI to create highly tailored, efficient solutions internally, or acquire specialized AI-native tools.
The evolution from chatbots to autonomous agents is a critical driver. These agents require sophisticated, vertically-specific data to learn and execute real-world tasks across diverse industries and functions. This marks a departure from simple data labeling towards creating intricate 'reinforcement learning (RL) environments' that simulate business workflows. Companies like Turing are at the forefront, generating this complex data to teach AI systems not just to pass tests, but to perform economically valuable work.
Navigating Enterprise AI Adoption: Opportunities and Constraints
While the technological capabilities of AI are advancing at a breakneck pace, enterprise adoption faces significant hurdles. Siddarth refers to these as the "first mile schlep" and "last mile schlep," encompassing challenges like fragmented internal data, the need for robust evaluation infrastructures, and designing workflows for partial autonomy (human-AI collaboration). Many incumbents, particularly in traditional sectors, struggle with internal processes and data quality, leading to a slow pace of implementation.
This inertia, however, creates immense opportunities. The speaker believes that value will inevitably transfer from slow-moving incumbents to agile, AI-native startups capable of embracing and deploying these transformative tools. While back-office automation might be a slower burn, front-office applications, especially in financial services, are poised for rapid adoption due to their direct impact on revenue generation and competitive alpha.
The Future of Work and Investment
Predictions about AI often lean towards dystopian scenarios of mass unemployment or a sudden 'rapid takeoff' of AGI. Siddarth offers a more optimistic view, advocating for a "slow and steady takeoff" of superintelligence. This gradual progression allows humanity time to adapt, reskill, and rethink education. He envisions a future where humans are 100x more productive, able to tackle multiple complex roles, and where entrepreneurship flourishes as non-technical founders can easily leverage AI to build companies.
In this evolving landscape, the investment thesis shifts dramatically. The immense prize of achieving AGI — which could dominate search, cloud, consumer devices, and even social networking — compels major tech players to invest hundreds of billions. This makes the cost of not participating in the AI race extraordinarily high. For investors, the focus must move beyond traditional SaaS to companies building defensible data-driven feedback loops, those specializing in custom AI deployments, and particularly emerging fields like robotics and embodied AI, which represent vast, untapped potential.
Conclusion
We are not in an AI bubble; rather, we are witnessing the early stages of a profound technological revolution. The models are powerful and improving daily, with a significant 'capability overhang' still to be unlocked through better integration and agentic scaffolding. The future promises an interconnected world where AI acts as an exoskeleton, amplifying human potential across all endeavors, solving humanity's grandest problems. Understanding these shifts is paramount for any business leader or investor aiming to thrive in the coming decade.
Action Items
Businesses should strategically invest in or partner with 'research accelerators' to secure high-quality, complex data for training their proprietary AI models and agents.
Impact: This proactive approach can significantly enhance AI model performance, create a competitive advantage through specialized intelligence, and accelerate the automation of core workflows.
Investors should reassess their portfolios for traditional SaaS companies, considering the potential disruption from foundation models and in-house AI development.
Impact: Reallocating capital towards companies focused on AI-native solutions, custom model deployment, or the 'first/last mile schlep' of enterprise AI integration may yield higher returns.
Enterprises must prioritize structuring and centralizing their internal data to overcome the 'first mile schlep' and effectively deploy custom AI solutions.
Impact: Improved data hygiene and infrastructure will enable more successful AI implementations, distilling proprietary human knowledge into models and unlocking significant operational efficiencies.
Leaders should adopt a hands-on approach, engaging directly with customer needs and ground truth, to navigate the rapid shifts in AI technology and market demands.
Impact: This leadership style fosters quicker adaptation, enables identification of critical areas for AI application, and ensures the development of truly impactful solutions for customers.
Governments should begin developing sovereign AI models and data collection strategies to maintain control over critical national infrastructure and sensitive information.
Impact: This ensures national security, promotes data privacy, and fosters independent AI capabilities, reducing reliance on foreign models and mitigating geopolitical risks.