Venture capital doesn’t concentrate like this by accident.
Two rounds. Two unicorn valuations. One 48-hour window. Cognition Labs, maker of the Devin AI software engineer, reportedly closed a $175M Series B at a $2B valuation led by Founders Fund, according to The Information. Perplexity reportedly closed a $62.7M Series B at a $1.04B valuation led by Daniel Gross, according to Bloomberg and Reuters. Both rounds carry verification caveats, source content could not be confirmed from available materials in this package, and human verification is required before publication. Both are presented throughout this piece with “reportedly” and qualified language reflecting that verification status.
The dollar amounts are not the story. The underlying thesis connecting both rounds is.
**The Category Signal**
What do an AI software engineer and an AI search engine have in common? At the product level, very little. At the investment thesis level, everything.
Both Cognition and Perplexity are betting on the same underlying premise: that AI can replace a category of knowledge work rather than augment a specific tool. Devin doesn’t help a developer write better code. It reportedly handles software engineering tasks autonomously. Perplexity Enterprise Pro doesn’t help a researcher use a search engine more effectively. It reportedly handles information synthesis and retrieval as a workflow replacement for traditional enterprise search.
This distinction, augmentation versus replacement at the category level, is what justifies unicorn-scale valuations for products that haven’t yet demonstrated enterprise revenue at scale. Investors aren’t pricing current performance. They’re pricing the total addressable market for replacing an entire job function with software that costs a fraction of the labor it displaces.
That thesis is large, speculative, and, if either company gets close to executing it, extraordinarily valuable. Founders Fund and Daniel Gross are making that bet explicitly.
**The Two Bets**
The specific bets are structurally different, which makes the co-occurrence more analytically useful than either round alone.
Cognition’s bet is vertical and high-stakes. Software engineering is a high-skill, high-compensation category. If Devin can reliably handle a meaningful fraction of an engineering team’s workload, not as a copilot but as an autonomous agent, the enterprise value of that capability is enormous. The risk is equally high: the gap between Devin’s controlled demonstrations and its real-world performance on complex codebases has been a documented and contested point since launch. Founders Fund’s $175M, at a $2B valuation, says they believe that gap will close.
Perplexity’s bet is horizontal and market-expansion-oriented. The enterprise search and knowledge management market is large, fragmented, and dominated by incumbents (Microsoft, Salesforce, ServiceNow) that have been integrating AI on top of existing architectures. Perplexity is building an AI-native alternative rather than an AI-enhanced legacy system. Enterprise AI has been outperforming consumer AI on revenue across recent cycles, Perplexity’s enterprise pivot is a bet that its brand recognition in consumer AI search translates to enterprise procurement.
| | Cognition Labs (Devin) | Perplexity AI | |—|—|—| | **Product type** | AI software engineer (autonomous) | AI search / knowledge management | | **Reported round** | $175M Series B | $62.7M Series B | | **Reported valuation** | $2B | $1.04B | | **Lead investor** | Founders Fund | Daniel Gross | | **Target buyer** | Engineering teams, CTOs | Knowledge workers, enterprise IT | | **Replacement thesis** | Software engineer labor | Enterprise search / research workflows | | **Verification status** | Qualified (sources broken/unverified) | Partial (primary URL broken; Bloomberg/Reuters cited) |
*All figures reported, human verification required before publication.*
**The Capital Pattern**
Who is investing matters as much as how much. Founders Fund’s track record includes early bets on Palantir, SpaceX, and Stripe, companies that built durable infrastructure for large markets. Their involvement in Cognition signals conviction that the AI software engineer category is infrastructure-level, not feature-level.
Daniel Gross is a former Apple and Y Combinator figure who has been actively investing in AI-native companies. His reported leadership of Perplexity’s round signals alignment between investor identity and company thesis, Gross has publicly articulated views about AI replacing knowledge work categories, not just improving them.
AI capital flows in 2026 have concentrated in two broad areas: AI infrastructure (hyperscaler commitments, chip companies, data center REITs) and AI application companies at the category-replacement layer. This cycle’s two rounds fall squarely in the second category. The infrastructure investment from Meta and Alphabet, covered in the companion brief, and the application investment in Cognition and Perplexity are two components of the same capital wave, serving different layers of the same stack.
Valuation multiples relative to revenue were not available in this package for either company. Revenue data for both Cognition and Perplexity is not publicly disclosed. The valuations are based on investor conviction about future revenue potential, not current multiples, a distinction enterprise buyers and analysts should hold clearly.
**The Enterprise Buyer Perspective**
If you’re a CTO, a head of engineering, or an enterprise technology procurement lead, these rounds present specific questions worth asking before any evaluation process begins.
For Cognition’s Devin:
– What does autonomous mean in your specific engineering environment? Devin’s capabilities are most relevant for well-scoped, modular tasks. Complex architectural decisions, security-critical code paths, and highly context-dependent work remain categories where human engineering judgment is not yet replaceable by documented external evaluations. – Has the product been evaluated on codebases like yours, not demo environments? Vendor-provided benchmarks for AI coding tools have historically shown a meaningful gap from real-world performance. Request independent evaluation data or run a structured internal pilot on representative tasks before any procurement decision. – What’s your liability posture on AI-generated code? Autonomous code generation at scale introduces questions about intellectual property, security review requirements, and compliance that procurement teams need to address before deployment, not after.
For Perplexity Enterprise Pro:
– What does “enterprise” actually include in the current product? Before any evaluation, confirm the specific enterprise features, data security, access controls, API integration, and audit logging, that distinguish Enterprise Pro from the consumer product. If those features aren’t fully built yet, the timeline matters. – What data leaves your environment when employees use the product? Enterprise AI search tools that query external knowledge bases introduce data governance questions. Understand the data handling architecture before onboarding sensitive workflows. – How does this fit alongside your existing Microsoft or Google ecosystem? If your organization is deeply integrated with Microsoft 365 or Google Workspace, the switching cost and integration complexity for an enterprise AI search alternative is a real procurement consideration, not just a feature comparison.
**What Comes Next**
Both companies’ rounds, if confirmed, will be followed by commercial pushes in the next two quarters. Series B capital at this scale is deployment capital, the companies will be hiring enterprise sales teams, building customer success infrastructure, and pursuing named enterprise accounts.
The shift toward production-grade agentic AI is accelerating in capital terms. The adjacent categories most likely to see similar investment concentration in the next 12–18 months include legal AI (contract review and drafting at the workflow level), financial analysis AI (earnings analysis, portfolio research), and enterprise content creation AI (marketing and communications at the category-replacement layer, not the individual-tool level). Each of those categories has the same structural profile as AI software engineering and AI search: a large existing labor market, high per-seat compensation, and AI capability that is approaching, but hasn’t yet reached, the threshold for reliable autonomous performance.
**TJS Synthesis**
Two rounds. Two theses. One shared premise: that AI is ready to be bought as a category replacement, not just a productivity feature.
Investors at Founders Fund and in Daniel Gross’s orbit are willing to price that premise at unicorn scale with limited verified revenue data. That is either prescient or premature, and the enterprise market’s response to Cognition’s and Perplexity’s commercial launches over the next 12 months will begin to answer which.
For enterprise buyers, the practical takeaway is simple: the products being funded at these valuations are real, they’re coming to your procurement pipeline, and the questions you ask before evaluation will determine whether you adopt a genuinely transformative tool or pay enterprise rates for a consumer product that wasn’t ready to cross the chasm. The investor conviction is established. The evidence of enterprise performance at scale has not yet arrived.
That gap, between investor thesis and enterprise proof, is where the next phase of this story lives.