The AI Industry Has a 50% Talent Shortage. Kleiner Perkins Is Spending $3.5 Billion to Fill the Gap Its Own Portfolio Created.

The $3.5B Bet: Kleiner Perkins Goes All-In on the AI 'Super Cycle'

On March 24, 2026, Kleiner Perkins closed $3.5 billion across two new funds: $1 billion for KP22, its 22nd early-stage vehicle, and $2.5 billion for KP Select IV, a growth-stage fund. The raise marked a 75% jump from the firm's $2 billion fund less than two years ago, and it signals something specific: Kleiner Perkins believes the biggest returns in AI won't come from the models themselves, but from everything built on top of them.

The firm's own language makes that clear. In a statement posted to its website, Kleiner called the current moment "one of the most important company-building moments in our lifetimes" and said the AI super-cycle is "still in the early innings." The firm sees AI expanding "across the full stack, from infrastructure to applications," with companies "taking root in the real economy faster than almost anything we have seen before."

That thesis is already visible in the portfolio. Kleiner holds early stakes in Together AI, Harvey, and OpenEvidence, plus positions in Anthropic and SpaceX (two companies expected to IPO this year). Over the past year, the firm has led or participated in increasingly large rounds: Applied Intuition's $600 million Series F, Chainguard's $356 million Series D, and Harvey's $300 million financing, according to VNTR's analysis. The sectors span autonomy, cybersecurity, legal tech, healthcare, and financial services, a deliberate spread across domains where AI adoption curves differ rather than a concentrated bet on any single category.

Kleiner is not alone in this positioning. Thrive Capital recently secured $10 billion in fresh commitments, and Founders Fund closed $6 billion for its fourth growth vehicle, as TechCrunch reported. But Kleiner's approach is structurally distinct: a lean team of five partners operating across both early and late stages, using early entry to secure long-term exposure while reserving substantial capital for follow-on rounds where ownership stakes are actually won or lost.

The firm's recent track record has given it room to be aggressive. Last year's IPO of Figma, a design software company whose $25 million Series B Kleiner led in 2018, delivered significant returns at a time when exits have been scarce. The firm also scored a return when its portfolio company Windsurf was acqui-hired by Google last summer.

What the new capital really buys is optionality across the AI stack. Kleiner's stated target areas (professional services, healthcare, autonomy, security, financial services, productivity, and the physical economy) map onto the layer of the market where AI stops being a research project and starts being a product. That is where the hiring surge begins.

What Kleiner Portfolio Companies Are Actually Hiring For

The new fund isn't chasing the next foundation model. Across roughly 279 portfolio companies, the roles being filled cluster around a specific thesis: the real money in AI is moving from training models to deploying them in production, and that requires an entirely different class of engineer.

At the infrastructure layer, companies like Nexthop AI ($500 million Series B, per VCBacked) are building networking hardware purpose-built for AI workloads, the kind of silicon-and-systems roles that demand firmware engineers, network architects, and compiler specialists. Together.ai, which runs an AI acceleration cloud, and Materialize, which builds a live data layer for AI agents, are recruiting the MLOps and distributed-systems engineers who keep inference pipelines running at scale.

Then there's the agentic orchestration layer, the software that lets AI systems act autonomously inside enterprises. Applied Compute, founded by former OpenAI reinforcement learning engineers and Scale AI data staff, builds custom enterprise AI agents trained on company data. Avoca is constructing what it calls an "AI workforce" for service businesses, handling inbound calls and lead management. Poetic turns plain-language runbooks into executable software. These companies need engineers who understand both large language models and the messy reality of enterprise workflows, roles that blend backend engineering with prompt design, tool integration, and reliability testing.

The security and trust segment is equally active. Armadin builds AI-powered offensive security tools, recruiting red-team operators who can work alongside autonomous agents. Gambit is constructing what it calls an "AI-native resilience platform," mapping infrastructure to find continuity gaps. Chainguard, focused on securing open-source supply chains, and Huntress, a managed cybersecurity platform, are both in growth-stage hiring. The common thread: these teams need security engineers who can audit AI-generated code, test autonomous systems for failure modes, and build the guardrails that let enterprises deploy agents without exposing themselves.

On the physical side, Applied Intuition, valued at $15 billion and working with 18 of the top 20 global automakers, is hiring across vehicle intelligence, autonomy, and defense. Mind Robotics, founded by Rivian's RJ Scaringe, is building physical AI for factory floors. Motive, the fleet-management platform, is scaling its AI-powered operations across trucking, construction, and energy. These roles lean heavily on robotics engineers, embedded systems developers, and computer vision specialists.

The healthcare vertical, which Kleiner has backed since co-founding Genentech in 1976, is now dominated by AI-native companies. Ambience Healthcare raised $243 million in a Series C to build an AI operating system for clinical documentation and coding. Hippocratic AI is developing safety-focused healthcare AI. Health Universe automates complex healthcare workflows. These companies need engineers who can navigate HIPAA constraints, integrate with legacy hospital systems, and build models that clinicians will actually trust.

What's conspicuously absent from this hiring map is the pure research scientist role, the kind of position that dominated AI recruiting in 2022 and 2023. Kleiner's portfolio companies aren't looking for people who can publish papers on novel architectures. They need engineers who can take a working model and make it reliable, fast, secure, and compliant inside a real organization.

This shift mirrors what's happening across the broader market. In Q2 2025, AI companies captured over 71% of all venture capital deployed, AInvest's flow analysis found. But that capital is concentrating in deployment and infrastructure, not in the next generation of foundation models. Kleiner's portfolio is a hiring blueprint for where the industry is heading: away from the lab and into the factory, the hospital, the fleet yard, and the enterprise back office.

The Exodus From Big Labs

The capital Kleiner raised in March 2026 is landing at a moment when senior AI talent is actively choosing smaller, faster-moving companies over the institutions that trained them. The firm's portfolio reads like a roster of the most aggressive hiring grounds in agentic AI: Harvey, Parallel, OpenEvidence, Glean, Hippocratic AI, Chainguard. All are scaling. All are recruiting.

The pull is structural, not just financial. At OpenAI, 38 roles were added on Zero G Talent's board in the past week alone, but they're spread across recruiting, business operations, campaign management, and forward deployment in Abu Dhabi. The research engineering roles that define the company's core technical mission sit alongside a growing layer of operational positions that look more like a scaled enterprise than a lab. Waymo's seven recent postings tilt similarly toward program management, technical writing, and post-silicon test engineering. The work is real, but it's specialized and deep inside large organizations.

What Kleiner-backed startups offer is a different equation. Periodic Labs, which raised a $300 million seed round in September 2025 led by Felicis and Andreessen Horowitz, was founded by former OpenAI and DeepMind researchers who wanted to automate science itself. They didn't join another lab. They started one with enough capital to hire aggressively from day one. The same dynamics played out at Thinking Machines Lab, which Mira Murati seeded at a $12 billion valuation in July 2025 with backing from Andreessen Horowitz, Nvidia, and Accel. These aren't companies poaching one or two senior researchers. They're building entire technical teams from the talent pools that OpenAI, Google, and Anthropic spent years developing.

Kleiner Perkins partner Mamoon Hamid said the firm is seeing "both the quality and quantity of founders and the ideas and opportunities at an all-time high." TechCrunch counted 55 US AI startups that raised $100 million or more in 2025, up from 49 in 2024, and eight of those companies raised multiple mega-rounds in a single year. The capital is there. The talent is moving. And the roles these companies are filling — agentic orchestration, AI security, firmware for autonomous systems, MLOps at scale — are the exact disciplines that big labs trained but can't always deploy fast enough.

Winston Weinberg, CEO of Harvey, the legal AI unicorn that Kleiner backed with a $300 million Series E in June 2025, said the firm's small partnership group and ex-operator background give it an edge in recruiting. "A lot of them are ex-operators, and that background is really helpful," Weinberg said. For engineers weighing a move, that pitch (direct partner engagement, faster decisions, less organizational friction) competes with the brand name of an OpenAI or a Waymo.

The risk, of course, is that the talent migration drains the labs that created the foundation models everyone is now building on. If enough senior researchers leave for growth-stage startups, the big labs slow down. But right now, the flow is one-directional, and Kleiner Perkins is positioning itself as the main beneficiary, backing the companies that catch the people the labs can't hold.

The Orchestration Layer: Why Making AI Do Things Is the Real Market

The AI workflow orchestration market hit $8.7 billion in 2024 and should reach $11.47 billion this year, according to MarketsandMarkets data cited by Growth HQ. By 2031, it's expected to blow past $35 billion. That growth isn't driven by foundation models getting better — it's driven by the messy, unglamorous problem of making AI systems actually do things inside real companies.

That's the orchestration layer: the software that lets autonomous AI agents plan, execute, and adapt across multi-step workflows with minimal human oversight. And it's where Kleiner Perkins is placing a significant portion of its newest bet.

The shift from copilots to coordinated digital teams is the story. A year or two ago, most enterprise AI looked like a chatbot bolted onto a helpdesk. Now, 79% of enterprises use AI agents in some capacity, and 96% plan to expand further in 2025, a LinkedIn analysis by Pankaj Tiwari found, drawing on McKinsey Global Institute and AIMultiple Research data. The global AI agent market nearly doubled from $3.7 billion in 2023 to $7.38 billion in 2025, and Gartner predicts that by the end of this year, 40% of enterprise workflows will include agentic components.

The orchestration layer is what makes that possible. Platforms like LangGraph, Microsoft AutoGen, CrewAI, and Relevance AI handle the graph-state management, multi-agent collaboration, and memory systems that let agents work together on complex goals (campaign optimization, supply chain simulations, compliance reporting) rather than sitting in isolation answering single questions. The LinkedIn analysis found organizations reporting 26–31% operational cost savings, with labor and integration costs dropping by up to 60–80% in some workflows.

Kleiner's thesis maps directly onto this shift. The firm's current investment focus prioritizes AI/ML infrastructure, agentic platforms, and autonomous systems, the production-grade plumbing that turns a demo into a deployed system. Its double-fund structure splits $1 billion for early-stage startups and $2.5 billion for growth-stage companies, a ratio that signals the firm is backing both the invention and the scaling of this infrastructure layer.

The talent demand follows the money. As orchestration moves from experimental pilots to governed, scaled deployment — Growth HQ reports that 50% of organizations will have developed AI orchestration platforms by next year — the engineering roles shift accordingly. It's no longer enough to fine-tune a model. Companies need people who can build modular agent networks, design event-driven architectures that connect ERP and CRM systems in real time, and embed compliance and audit trails into every workflow. Nearly all large organizations still cite fragmented systems as a major integration barrier, which means the engineers who can stitch together legacy and cloud environments are the ones getting hired.

The bottleneck is no longer intelligence — it's integration. Models are getting cheaper and more capable at a pace that dwarfs the industry's ability to deploy them in production. Kleiner Perkins is betting that the next trillion-dollar companies won't be the ones with the best model, but the ones with the best orchestration — the factory floor where autonomous agents are composed, governed, and scaled. The hiring surge its portfolio companies are running reflects that bet in real time.

Where the Jobs Actually Are

The Bay Area still eats the Kleiner-backed AI hiring surge whole. LinkedIn data shows 126 open positions tied to Kleiner companies in the Palo Alto radius, with another 174 listed on Glassdoor for San Francisco proper. But the sub-geography inside that concentration tells the real story about what these startups are building.

San Mateo has emerged as the single densest hiring corridor. Maxima, the AI math-reasoning company, lists roles almost exclusively there: software engineers, frontend engineers, data infrastructure leads, BD reps, all at San Mateo addresses. Dexterity, the robotics firm, clusters in Redwood City. Hippocratic AI splits between Menlo Park and Palo Alto. MAI Agents and Inworld AI both pull from Mountain View. Parallel Web Systems and Labelbox list roles across the broader Bay Area without pinning a single city.

The pattern is visible in the table below, built from live LinkedIn job data filtered around the Palo Alto hub:

What jumps out is the on-site mandate. Of those 126 Kleiner-linked roles around Palo Alto, 105 are on-site, 20 hybrid, and just one remote. These companies are hiring people who will sit in the same building where the hardware is tested or the agent orchestration stack is being debugged. Dexterity has a materials handler opening alongside its robotics systems test engineers in Redwood City, the factory floor and the engineering floor share a zip code.

The secondary markets are real but thin. Day One Partners posted a software engineer role for a Kleiner-backed AI startup in New York. Handshake, which Kleiner backs, is San Francisco-based and describes itself as a career network for the AI economy. Echo AI lists roles through Kleiner's jobs board. None of these secondary hubs approach the Peninsula's density.

For engineers weighing relocation, the signal is blunt: if you want the widest set of Kleiner-backed AI infrastructure roles in one commute, aim at the San Mateo–Redwood City–Mountain View corridor. The companies building the orchestration layer and the physical-systems layer want you within walking distance of the prototypes.

What the Roles Reveal About Skills That Actually Matter

Strip away the company names and funding stages, and a clear pattern emerges: the skills these startups are hiring for have almost nothing to do with training foundation models and everything to do with making AI systems work in production.

LinkedIn's job board for Kleiner-backed companies lists over 1,300 open roles in the United States alone. Software engineer positions dominate, but drill into the specifics and the picture gets more interesting. Rippling, a payroll and HR platform, is hiring backend-focused software engineering interns, machine learning software engineers, and full-stack engineers across San Francisco, Seattle, and New York. Secureframe, a compliance automation company, lists roles for new grad software engineers in both growth and product tracks. MAI Agents, based in Mountain View, is looking for full-stack engineers and machine learning engineers. Maxima, in San Mateo, has open roles for frontend and backend software engineers.

The common thread isn't model development. It's infrastructure engineering, the plumbing that connects AI capabilities to real business workflows.

This tracks with what's happening at the frontier labs themselves. OpenAI recently posted a role for a "Software Engineer, Agent Infrastructure," describing the position as building "systems to train highly capable agentic models" and "the platform and integrations to launch new agents to hundreds of millions of users." The job sits at the intersection of research and production, exactly the zone where Kleiner's portfolio companies operate.

Meanwhile, the non-engineering roles reveal another dimension. Profound, an AI-powered market intelligence platform, is hiring visual designers and product designers. Synthesia, the video generation company, has openings for digital designers and technical support associates. Stord, a logistics platform, lists associate product manager and associate data roles for new grads. These aren't research positions. They're the roles you create when a product has moved past the prototype phase and needs to ship, support, and sell.

Zero G Talent's own board data reinforces the pattern. OpenAI added 38 roles in the past week, including a Forward Deployed Engineer in Abu Dhabi and a Business Operations Partner for its Transportation Program, positions that have nothing to do with model training and everything to do with putting AI into specific industry contexts. Waymo posted seven roles, several in silicon validation and post-silicon testing, which are fundamentally hardware-infrastructure jobs masquerading under an AI company's name.

For engineers navigating this market, the signal is sharp: the premium has shifted. Deep learning research skills still command high salaries, but the volume (and the velocity of hiring) is concentrated in orchestration, MLOps, agent infrastructure, and the full-stack engineering required to wrap models in usable products. If you're deciding where to invest your learning time, the job boards of Kleiner's portfolio are a more honest signal than any keynote.

How Kleiner's Hiring Blitz Fits Into the Industry's Pivot From Research to Production

The Kleiner Perkins fund isn't just a bet on AI — it's a bet on a specific phase of the AI lifecycle that most of the industry is only now waking up to. The era of pouring capital into foundation model research is giving way to something less glamorous and far more labor-intensive: making AI systems actually work in production. Kleiner's hiring surge maps directly onto that shift.

IDC's 2025 AI-Fueled Organization Maturity Model puts the transition in stark terms. In 2023–2024, most firms were in the "Ad Hoc" or "Opportunistic" stages, running scattered proofs of concept with no central AI strategy. By 2025, IDC says the industry has entered the "AI Pivot" phase, where organizations move from experimentation to repeatable, production-grade deployment. The bottleneck is no longer model quality. It's everything around the model: governance, data pipelines, monitoring, orchestration, security, and the infrastructure that lets an AI system run reliably at scale for months or years without human babysitting.

That's exactly the layer Kleiner is hiring into.

Goldman Sachs offers a concrete example of what this pivot looks like from the demand side. The bank rolled out its GS AI assistant to roughly 10,000 employees in early 2025, with plans to cover all knowledge workers that year. Chief Information Officer Marco Argenti described the tool's evolution in three stages: first, answering queries based on internal data; second, exhibiting "agentic behavior" by completing multistep tasks on behalf of employees; third (three to five years out), reasoning through problems the way a seasoned Goldman analyst would, without being handed a runbook. JPMorgan Chase and Morgan Stanley have made similar pushes, with more than 200,000 and roughly 40,000 employees, respectively, given access to in-house generative AI tools.

The pattern is consistent: enterprises aren't hiring more researchers to build better models. They're hiring engineers to deploy, secure, and maintain the models they already have.

Magnit Global's workforce data backs this up. In Q1 2025, AI/automation role fills doubled year over year, rising from 3% to 6% of total fills. But the composition shifted dramatically. Data engineering roles, the backbone of training and experimentation, dropped from 46% to 32% of AI/automation fills. Automation roles, which cover the orchestration and operational side, jumped from 32% to 44%. Overall IT/tech fills actually contracted 2% year over year. The growth is concentrated in production, not research.

LinkedIn's global job market analysis tells the same story at scale. By 2025, an estimated 75% of enterprises will have moved AI models into full production environments. That transition is driving what it calls "skyrocketing demand" for MLOps engineers, specialists who build deployment pipelines, automate model updates, and maintain AI infrastructure. Job postings requiring AI skills grew second-fastest in cybersecurity roles, up about 25%, reflecting the new threat surface that production AI systems create.

The talent shortage is acute. LinkedIn's analysis cites a survey finding that 76% of large companies reported a severe shortage of AI talent on their teams, even as 93% viewed AI as critical to their future. India, which has one of the world's largest AI workforces at roughly 600,000 professionals, may only be able to fulfill about half of anticipated AI talent demand, a 50–60% supply gap by mid-decade.

This is the gap Kleiner Perkins is trying to fill. The firm's portfolio companies aren't competing with OpenAI for the same handful of transformer researchers. They're hiring the engineers who build the orchestration layers, the security frameworks, the deployment tooling, and the agentic systems that turn a demo into a product. The new fund is, in effect, a wager that the next trillion-dollar AI companies won't be model makers, they'll be the ones that make models usable.

A LinkedIn analysis of job ads found that listings requiring AI skills offer 25% or more in salary premiums over comparable roles that don't. But the roles carrying that premium are increasingly in deployment, operations, and security, not in training the next generation of large language models. Kleiner Perkins wrote the biggest check to back that bet. The 1,300-plus open roles across its portfolio are the proof.

Working in AI? Zero G Talent tracks the openings: browse AI jobs, openings at OpenAI, Waymo and Super, and the people building the field.