Aug 11, 2026 · 38m · big-technology
AI Agents Are Creating A Data Explosion. Here's What To Do About It. — With Clint Sharp
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Cribl CEO Clint Sharp joins Alex Kantrowitz to discuss how autonomous AI agents are triggering an explosion in enterprise telemetry data and reshaping cybersecurity. Sharp outlines the necessity of real-time stream processing, AI observability, and open security access to navigate this technological transition.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Alex holds 22.9% of the talking time here. How this is scored →
Alex as informed peer 5.0 Guest teaching 5.7 Guest disagreement 2.3 Alex pushing back 2.3
speaking balance: gold is Alex, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 22:55 Sharp denounces Anthropic safety rhetoric as fear-mongering
Sharp forcefully attacks Anthropic's public warnings as fear-mongering, arguing that comparing AI models to nuclear weapons only serves to justify harmful government intervention.
Hardest push from Alex ▶ 24:10 Kantrowitz challenges Sharp on dismissing frontier AI safety concernsKantrowitz refuses Sharp's framing that Mythos was pure marketing hype, pressing him on whether models discovering zero-day vulnerabilities in facts justify rigorous alarm and guardrails.
Biggest teaching moment ▶ 22:50 Sharp reveals defenders resorted to Chinese models due to Western safety refusalsSharp educates Kantrowitz on the unintended consequences of Western model guardrails, explaining that Hugging Face defenders had to switch to open-source Chinese models to inspect their own security vulnerabilities.
Alex holds their own ▶ 3:03 Kantrowitz details Nadella's reverse information paradox thesisKantrowitz demonstrates detailed knowledge by quoting and breaking down Satya Nadella's thesis regarding how model providers capture competitor know-how through user corrections and tool traces.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Alex as informed peer | Guest teaching | Guest disagreement | Alex pushing back | Why |
|---|---|---|---|---|---|---|
| The Reverse Information Paradox and Demand for AI Observability | 6 | 4 | 1 | 2 | Kantrowitz displays strong context by quoting Satya Nadella's memo on the reverse information paradox. Sharp validates the premise and explains how enterprise demand for AI observability emerged almost overnight. | |
| Assessing the Economic Value and Costs of AI Exhaust | 5 | 6 | 2 | 3 | Kantrowitz cites Alex Karp's warning about subsidized AI models harvesting proprietary telemetry. Sharp pushes back on Karp's theory, explaining that frontier models are expensive and naive exhaust storage is economically non-viable. | |
| Autonomous Agents and the Acceleration of Data Generation | 5 | 6 | 1 | 1 | Kantrowitz brings up Greg Brockman's estimates regarding agent adoption. Sharp explains how machine-speed autonomous agents generate 3-5x the telemetry volume of human users across enterprise systems. | |
| Autonomous Security Threats and Enterprise AI Deployment Obstacles | 5 | 6 | 3 | 2 | Kantrowitz brings up the OpenAI-Hugging Face breach incident to ask about AI security. Sharp explains the dual risk of automated attacks versus split-brain operational outages caused by hallucinated defenses, while venting about hypocritical CISO procurement hurdles. | |
| Debating AI Safety Guardrails, Anthropic Mythos, and Cyber Defense | 6 | 7 | 6 | 5 | Sharp strongly criticizes Anthropic's fear-mongering and safety guardrails, revealing that Hugging Face defenders had to use Chinese models. Kantrowitz pushes back, asking whether frontier model capabilities like autonomous zero-day discovery actually warrant serious alarm. | |
| Cribl's Data Architecture and Managing Expanding Telemetry Costs | 5 | 6 | 1 | 2 | Kantrowitz introduces the Wikipedia scraping traffic spike as an analogy for agent data loads. Sharp elaborates on how agents without pain thresholds explore multiple hypotheses simultaneously, requiring new GPU wire-rate filtering architectures. | |
| Software Commoditization, Customer Trust, and the AI Frontier Outlook | 3 | 5 | 2 | 1 | Kantrowitz asks whether escalating AI infrastructure costs will ultimately pay off. Sharp outlines his thesis on the complete commoditization of software features via AI, arguing enterprise value will shift entirely to customer trust and fair pricing. |
Statements from this episode (16)
Kantrowitz: Cribl Has 1,200 Employees and $350 Million in ARR
“It's a nine year old company, 1200 employees, three hundred fifty million ARR”
Sharp: Cribl's Largest Customers Move Petabytes of Telemetry Data Daily
“Our largest customers move petabytes of information daily, whereas like a data warehouse, as an example, like that might be, ah, a petabyte data warehouse is a huge data warehouse. Like that's for like a Snowflake or a Databricks, like that's a big data wareho…”
Sharp: AI model providers will launch competing telemetry products within 18 months
“I have a near 100% belief that the, that these companies will eventually want to compete with my business. And so, you know, I need them. I need their models. I need the intelligence that they're selling, but I'm, I have some very real concerns that, you know,…”
Sharp: A centuries-old enterprise is deploying Claude to 50,000 desktops
“I have a customer, you know, a normal run-of-the-mill brand that you would know, not a company that you would think is like an innovative company, like it's multiple hundreds of years old. They're rolling out Clawed to 50,000 desktops.”
Sharp: Individual AI traces are worthless, but aggregated data is immensely valuable
“Individually looking at one user's trace is essentially worthless, unless I have some security problem I need to go urgently look at it, but normally it's worth nothing. But in aggregate, it's incredibly valuable”
Sharp: OpenAI and Anthropic are not underpricing tokens to harvest data
“OpenAI and Anthropic in particular they're not cheap. Not by comparison to what we're looking at with OpenWeights models, etc. So, I don't believe that they are possibly subsidizing You know, token, their premium tokens in order to get this reinforcement learn…”
Sharp: Most enterprises are years away from sophisticated AI model improvement
“And kind of where I see most enterprises at, I think they are likely
Years away from being able to do anything that, that sophisticated.”
Sharp: Frontier AI models are more skilled than the average cyberattacker
“Like, the, what I have available in K-III or GLM-VII is better, like, is probably more skilled than the average attacker was before, and they're going to move super, super fast.”
Sharp: Enterprise resistance to vendor AI terms will vanish in five years
“It's getting better, and that's a temporary, I mean, in five years we'll have forgotten that, that this was even a thing, but it is definitely an impediment”
Sharp: Hugging Face used open-source Chinese AI to defend an OpenAI jailbreak
“My CISO talked to, as a group of CISOs, went and talked to the defending team at Hugging Face, and this absolutely happened. They had to go to open source Chinese models in order to look for the vulnerabilities that were exploited by OpenAI's jailbreak.”
Sharp: 8 to 10 Frontier AI Labs Are Building Equivalent Cyber Capabilities
“The reality is, is that the, this same weapon that's being invented is being invented by eight to 10 different frontier organizations all at the same time. And so you can't non-proliferate when that many people are building the same thing at the same time.”
Sharp: Cybersecurity faces a severe 12-month crisis before AI defense rebounds
“We're gonna have a really bad year. In terms of security, ah, and hacking in general, and people, ah, you know, being exploited. It's gonna be a bad year. But we only really have to go through it once. As the cost of this intelligence comes down, we will event…”
Sharp: AI agents multiply data queries because they lack human friction
“If I put an AI agent on top of asking questions of that data the agent's going to ask a lot more questions than any human ever did. The, they move at machine speed, and so they explore all of these hypotheses, and they go down all of these different roads, and…”
Sharp: Cribl ships 2x to 3x more software than a year ago
“We're seeing this with our own engineers now, like the amount of software that we're shipping at Kribble, I mean, we're shipping in a factor of two or three what we were a year ago. I mean amazing productivity increases.”
Sharp: AI will commoditize software features across all vendors
“Building software is so productive, especially copying software. So if I, like, if you have a feature, I can implement a version of that feature. Super easy. You point the models at it, and they make you something that looks like that, feels like that, does th…”