Turing CEO, Jonathan Siddharth: Who Wins in Data Labelling & Why SaaS Will Die?

1. Key Themes

The Evolution from Data Labeling to Research Acceleration

The fundamental shift in AI data needs has transformed Turing from what was perceived as a "talent marketplace" into a research accelerator for frontier AI labs. Jonathan explains: "I think the era of data labeling companies is over and it's now the era of research accelerators. The labs want to work with a proactive partner that can think about what types of data are likely to be helpful for these models and can make recommendations to them" [00:12:14]. This transformation is driven by three major shifts: data complexity (from simple tasks to sophisticated workflows), focus on real-world work (not just passing tests), and the transition from chatbots to agents requiring reinforcement learning environments.

The $30 Trillion Knowledge Work Automation Opportunity

Jonathan presents a compelling vision that all digital knowledge work will be automated, describing it as a four-dimensional matrix: "The first dimension is every industry like financial services retail, healthcare, podcasting. The second dimension could be every function software engineering, marketing sales, finance. The third dimension could be every role in that org chart. The fourth dimension is a workflow that a human goes through in that role. We are creating oral environments for every workflow, for every role in every function in every industry. That's like $30 trillion of knowledge work" [00:08:05]. This comprehensive approach to data generation positions Turing uniquely in understanding where AI models break in reality.

The Death of Traditional SaaS

Jonathan boldly declares the end of SaaS as we know it, driven by three converging forces. "SaaS as we know it, I think is over. I feel like quite a few SaaS apps were built at a time when software was relatively hard to build and complex to build" [00:52:30]. The risks include: companies building custom software easily themselves, getting "Sonic boomed" by foundation model companies moving into applications, and the obsolescence of GUI-based software designed for human keyboard-and-mouse interaction. He argues that in an agentic AI world, ambient AI interfaces will replace traditional software navigation.

2. Contrarian Perspectives

Slow Takeoff Theory Benefits Humanity

Contrary to popular rapid AGI predictions, Jonathan advocates for gradual AI progress: "I don't think we'll see rapid takeoff. I think we'll see incremental continuous improvement in AI. I actually think this is good for the world because if what we believe happens, which is all types of digital knowledge work gets automated. I think humanity needs time to prepare its workflows" [01:08:05]. He distinguishes AI from self-driving cars, arguing that incremental improvements unlock proportional value rather than requiring 100% accuracy for utility. This measured approach allows for workforce adaptation and continuous value realization.

Front Office Adoption Will Outpace Back Office

Against conventional wisdom about enterprise back-office automation, Jonathan argues: "My hypothesis is that companies will be very slow with back office automation, but in the front office, for example, I speak with financial services clients in New York, like some of the biggest companies...they are extremely interested in applying AI to help them make better investment decisions because it directly translates into helping them make more money. I've found it's a lot easier to convince people to use a piece of technology to make more money than to save money" [01:09:08]. This challenges the assumption that cost-saving automation will drive initial AI adoption.

More Software Engineers, Not Fewer

Contradicting fears of engineering job displacement, Jonathan predicts: "I think the definition of a software engineer will change. A Stanford doctor who's in oncology who has an idea for like some cancer detection type app. Now that person will be able to create a very simple version of an app...I think there'll be more software engineers because if you define a software engineer as somebody who's capable of building a software product to solve a real problem, that pool of builders is going to expand way beyond people who've graduated with like a four year computer science degree" [00:58:44]. This reframes AI not as replacement but as democratization of technical capability.

3. Companies Identified

Scale AI

Description: Data labeling and AI infrastructure company, competitor in the AI data space

Why Mentioned: Jonathan specifically called out Alex Wang and Scale as worthy of respect for their prescience and ability to navigate market shifts. "I have a lot of respect for Alex Wang from Scale AI. I feel like Alex and Scale were prescient in seeing the importance of data. And I admired how having started in autonomous labeling, like how they kind of navigated the ups and downs" [00:38:29]. He noted Scale's strength in multi-modality due to their autonomous vehicle heritage and that their acquisition created massive demand spillover for Turing.

OpenAI

Description: Leading AI research laboratory and frontier model developer

Why Mentioned: Cited as research partner and for their GDP Value paper measuring AI's economic impact. "I encourage your listeners to look up GDP value, which is this paper by OpenAI, where they measured the impact of today's AI models in automating all types of economically valuable work...they saw that today's models were quite good at achieving parity with the best human experts in that field" [00:22:17]. Their research demonstrates that current AI achieves 50% parity with human experts on single-step tasks.

NVIDIA

Description: GPU manufacturer and AI compute infrastructure provider

Why Mentioned: Used as comparison for revenue concentration concerns. "The last time I checked, like I was told that NVIDIA for NVIDIA, 39% of their revenue comes from two clients. And roughly 50% was like four clients" [00:41:23]. This validates Turing's similar customer concentration as normal for frontier AI infrastructure businesses.

Cursor

Description: AI-powered code editor

Why Mentioned: Exemplar of effective partial autonomy design. "Andre Karpati articulated this first. When talking about why cursor works so well, because it's not designed for full autonomy, it's designed today for partial autonomy for humans to collaborate with the AI to do that specific task. So that cursor for X needs to be built for every role, for every workflow" [00:47:34].

4. People Identified

Alex Wang

Description: Founder and CEO of Scale AI

Why Mentioned: Specifically praised for leadership and operational excellence. "I have a lot of respect for Alex Wang from Scale AI...I really like the way he operates as well. I feel like there are certain elements of leadership that I think I share with him" [00:38:41]. Jonathan admires how Alex navigated Scale through market transitions and saw the importance of data early.

Elon Musk

Description: Entrepreneur, CEO of multiple companies including xAI

Why Mentioned: Cited as operational role model and example of AI safety consciousness. "One of my learnings from Elon's biography is that he was so hands-on. He would be walking the factory floor and asking an engineer why this door in the Model 3 has three bolts instead of maybe two" [01:13:31]. Also defended Elon's approach to AI safety: "his motivation for getting into AI was that he wanted an AI that was species and loves humanity" [01:11:53].

Sam Altman

Description: CEO of OpenAI

Why Mentioned: Named alongside Elon Musk when asked which leaders Jonathan most respects in the AI space [00:38:18].

Andre Karpathy

Description: AI researcher and former Tesla AI director

Why Mentioned: Credited with articulating why partial autonomy works. "Andre Karpati articulated this first. When talking about why cursor works so well, because it's not designed for full autonomy" [00:47:34].

Mark Chen

Description: Head of Research at OpenAI

Why Mentioned: Cited on financial services adoption patterns. "I've heard Mark Chen, the head of research at OpenAI say this about how financial services is usually at the bleeding edge among all the other industries in the S&P 500. But even they are like about two years behind usually, like relative to the state of the art" [00:19:58].

5. Operating Insights

The Tandem Training System for AI Deployment

Jonathan reveals Turing's deployment methodology: "The way, for example, we do deployments is like a tandem system where you'd have a human and an AI doing the same job for a period of time, where a manager can see the output of both. If the agent is right and the human is wrong, you train the human. If the human is right and the agent is wrong, you've created a data point to fine tune the next iteration of the agent. So the agent is steadily improving over time" [00:33:02]. This creates a continuous improvement loop while maintaining quality control.

First Mile and Last Mile Schlep

Jonathan identifies often-overlooked implementation challenges: "First-mile schlep is...our data is a mess. It's in silos. It's super fragmented. Some of the data is in spreadsheets. Some of the data is in a file that Bob has. And Bob doesn't work here anymore. You first have to acquire the data, convert the unstructured data into structured data into a format to find tune LLMs. You might want to set up good infrastructure for e-vals" [00:31:40]. This explains why 95% of AI pilots fail - not due to model limitations but implementation complexity.

Abandoning Leverage for Ground Truth Proximity

Jonathan shares a fundamental operating philosophy shift: "I used to believe that to build a enduring valuable company, you hire a strong exact team and operate with a lot of leverage. Basically, hire strong people and get out of the way. I used to believe that. Now I believe you hire great people and work really closely with them and their directs and their directs and get as close to the ground as you can, where ground truth usually exists with the customers" [01:12:40]. This anti-delegation approach, inspired by Elon Musk, prioritizes deep involvement in critical details.

Data-Driven Feedback Loops as Moat

When asked about moats in an AI world, Jonathan emphasized: "I think one mode will be data-driven feedback loops. For example, one reason Google had such a great lead in search for a while was these data-driven feedback loops that come from people using your product and generating data that gives you the algorithm developer a high quality gradient for which direction to step in" [00:29:18]. The advantage isn't the algorithm but the quality of improvement signals from real-world deployment.

6. Overlooked Insights

Revenue Concentration as Feature, Not Bug in AI Infrastructure

While Harry pressed on revenue concentration risk (working with only 7-8 frontier labs), Jonathan made a remarkable comparison that went largely unexplored: "The last time I checked, like I was told that NVIDIA for NVIDIA, 39% of their revenue comes from two clients. And roughly 50% was like four clients" [00:41:23] for a $5 trillion company. This suggests AI infrastructure may inherently concentrate around frontier labs, similar to how compute concentrates, making this a permanent market structure rather than a transitional risk. The implication is that investors should evaluate AI infrastructure companies by different metrics than traditional B2B software.

Model Capability Overhang Creates Immediate Opportunity

Jonathan briefly mentioned but didn't fully develop: "There is a very significant model capability overhang. By that, what I mean is the models are capable of X but what we are getting out of the models is X minus delta. There is with the right agentics scaffold around these models in terms of the right system prompts, the right user prompts giving the models access to the right context, teaching the models how to acquire additional context, teaching the models how to use the right internal tools. There is significant amount of capability that can be unlocked with today's models" [00:44:55]. This suggests the biggest near-term opportunity isn't waiting for better models but better deployment infrastructure - a multi-billion dollar implementation gap that exists today with current technology.