Nutanix's Aug 4 Cut Protects NKP. The Real Pool Is ~180.

On August 4, 2026, Nutanix filed an 8-K disclosing a ~5% workforce cut, roughly 390 people, to be substantially complete by end of October. Every infra recruiter I know immediately started blasting "Nutanix" LinkedIn profiles. Most of them are sourcing the wrong pool.

The CEO's own language names the teams that are protected. Read it carefully and the addressable pool shrinks from 390 to something closer to 180, most of them not in the Bay Area.

What the Aug 4 8-K actually says

Nutanix announced a ~5% cut with $33M to $43M in pre-tax charges, substantially complete by end of October 2026, but explicitly protected AI, NKP, external storage, and customer-facing sales from the reductions. This is a reallocation cut, not a distress cut, and that framing changes how you source and how you pitch.

The numbers from the filing and the earnings context:

• 8-K filed Aug 4, 2026, disclosing a workforce reduction of approximately 5% of global headcount.
• Approximately 390 employees affected, derived from 5% of Nutanix's ~7,800 headcount.
• $33M to $43M in aggregate pre-tax charges, primarily one-time severance and termination benefits, a substantial majority in future cash expenditures.
• Implied severance of ~$85K to $110K per head, consistent with a US-heavy, senior IC and manager mix.
• Q3 FY26 revenue of $703M, up 10% year over year. ARR of $2.43B, up 15%. More than 700 new customers added in the quarter.
• FY26 full-year ARR closed at $2.55B, up 16% YoY, on Q4 revenue of $757.1M.
• Scope and timing vary by jurisdiction, subject to works council and consultation processes.

A Nutanix spokesperson named the protected investment areas directly: "AI, modern application platforms with NKP, infrastructure modernisation through external storage, and customer-facing sales resources." That sentence is doing all the work. It tells you which teams the ~390 came from (legacy HCI, AHV/Prism maintenance, non-NKP platform tooling, G&A, support) and which are off-limits.

$33M - $43M

The pool is not 390, it is closer to 180

The addressable pool for a K8s or infra startup is not the full 390 layoff, it is the narrow slice of NKP-adjacent engineers whose specific team got cut. In Refolk's index of professional profiles, only ~30 profiles explicitly reference NKP or Kubernetes work tied to Nutanix, out of 233 total Nutanix engineering profiles. Scale that signal to full LinkedIn coverage and the company-wide NKP-adjacent engineering population is on the order of 180 people.

Now overlay the CEO's carve-out. NKP is protected. The Karbon veterans who moved onto the NKP brand are (mostly) staying. The engineers who are actually on the market are:

1. Legacy Karbon and pre-NKP Kubernetes tooling engineers whose scope got folded.
2. AHV and Prism platform engineers on maintenance-mode teams.
3. HCI QA, release engineering, and support staff.
4. G&A, technical program managers, and non-NKP infrastructure PMs.

Sourcing "Nutanix + Kubernetes" gets you the protected group. Sourcing "Nutanix + Kubernetes + Karbon" or "Nutanix + AHV + platform" gets you closer to the people who are actually calling recruiters back this month. Describing that intent in plain English is the exact gap Refolk closes: ask for "ex-Nutanix engineers on Karbon or legacy AHV platform teams, not current NKP" and you get a ranked shortlist instead of 233 undifferentiated profiles.

The table nobody is publishing

Here is the addressable-pool math, laid out. Every number below is either from the SEC filing, from Nutanix's public reports, or from Refolk's index of Nutanix engineering profiles.

A founder who thinks the pool is 390 will price offers as if candidates are competing for scarce seats. A founder who understands the pool is ~180, and that only a fraction of those are actually exiting, will price offers as if they are competing with Rubrik and Pure Storage. The second founder wins.

Bengaluru is where the volume is, not San Jose

If you are running US-focused Nutanix sourcing, you are fishing in the smaller pond. Refolk's index shows roughly 4x more Nutanix engineers in Bengaluru than in San Jose, matching Nutanix's known dual-hub R&D structure. Bengaluru alone is ~32% of the engineering pool in the sample.

Two mechanics make this the real hiring race:

• Salary arbitrage. Indian startup salaries run 30% to 50% below Nutanix's, which sounds like a founder's dream until you realize Rubrik and Cohesity are bidding in the same Karnataka market with cash and stock. If you are a US Series A pitching remote India roles, you are competing with Bengaluru-native compensation, not with your NYC comp band.
• Alumni network access. Informal ex-Nutanix Bengaluru WhatsApp and LinkedIn groups exist that Bay Area sourcers do not know about. Warm intros through Rubrik alumni (Bipul Sinha's network) and Cohesity alumni (Mohit Aron's network) route through Bengaluru first, San Jose second.

Bengaluru also skews toward AHV and Prism veterans, exactly the profile most likely to be exiting under the CEO's carve-out. If your product needs someone who has actually shipped a hypervisor storage stack, that is the geography.

The Oct 31 window is a US illusion

The "substantially complete by end of October" language applies to the aggregate, but the 8-K explicitly flags that scope and timing vary by jurisdiction and are subject to works council and consultation processes. EU exits will trail the US window by weeks or months.

Practical implications for anyone sourcing infra engineers in Q4:

• US roles (San Jose, Durham, remote US) close by Oct 31, 2026. If you want to close before Rubrik and Pure Storage, your outreach needs to land by mid-September.
• France, Germany, Netherlands exits likely finalize between November 2026 and January 2027. Munich, Paris, and Amsterdam are a hidden second window.
• India exits typically move fastest of any region, often ahead of the aggregate deadline, because there is no works council delay. Bengaluru will feel like a spike in September, not October.

The founders and recruiters who understand these three timelines can run three separate sequences instead of one panicked October blast. A recruiter who shows up in Munich on November 5 with a real role is competing against almost nobody.

A recruiter who shows up in Munich on November 5 with a real role is competing against almost nobody.

Severance of ~$90K means candidates are not desperate

Because implied severance sits at ~$85K to $110K per head, most of these engineers can afford to wait three to six months for the right role, which changes both your pitch and your pricing. This is not a WARN-notice fire sale. It is a negotiated reallocation, backed by a company that just closed FY26 with $2.55B ARR up 16% YoY.

What that means for your outreach:

1. Skip the "we move fast" pitch. These candidates have runway. Speed is not the hook.
2. Lead with technical scope, not equity upside. An AHV veteran who spent seven years on a hypervisor does not want to write a CRUD API for your seed-stage startup. Show them the storage layer or the K8s control plane you are actually building.
3. Expect Rubrik and Cohesity to close the November wave. If your comp is more than 15% below theirs on cash, you will lose the exits who wait past the US October window.
4. Reference the alumni network by name. Mentioning that your CTO worked with someone at Cohesity or Rubrik is signal, not name-dropping. This community is small.

The candidates who reply fastest are the ones from adjacent-but-non-K8s teams (HCI QA, release engineering) who realize their titles do not translate cleanly and want to reposition. They are your fastest closes. The NKP Kubernetes hiring race is a slower, more expensive game.

The four signals that separate the pool

Search on the actual project keywords, not the marketing brand, because almost nobody self-describes as an "NKP engineer" externally. Refolk's index shows the term "NKP" barely appears in headlines or summaries even among Nutanix engineers.

The signals that actually surface the right people:

• Karbon (the pre-NKP name, still on many résumés).
• AHV and Prism (the hypervisor and management plane, high concentration of platform veterans).
• Kanister and D2iQ references (the acquired-tech lineage inside NKP).
• Adjacent employer history: Google, Microsoft, Meta, Rubrik, and New Relic are the top prior and next-move companies in Refolk's Nutanix engineering pool. Rubrik in particular is a natural landing pad and a natural poach target.

Running these searches manually across LinkedIn and GitHub takes a week to build the query and still misses the alumni-network overlap. Describing the shape of the person in plain English ("ex-Nutanix Karbon or AHV, now open, Bengaluru or EU, with Kubernetes control-plane experience") to Refolk collapses that week into a ranked list, cross-referenced against GitHub commit history so you can see who actually shipped versus who managed.

What to do this week

If you are hiring against this window, the sequence is: identify the ~180-person NKP-adjacent pool, exclude the protected NKP team, prioritize by geography-timed exit, then pitch on scope. Concretely:

1. Build the Bengaluru list first. It is the largest, moves fastest, and has the least US-recruiter competition.
2. Build the France, Germany, Netherlands list second. Their window opens as the US window closes.
3. Reserve the US Bay Area list for last. Rubrik and Pure Storage will out-cash you here unless you have Series C money.
4. Ignore anyone whose headline currently says "NKP" or "AI Platform" at Nutanix. They are the protected group.
5. Prioritize Karbon, AHV, Prism, HCI QA, and D2iQ-lineage keywords over generic "Kubernetes."

The Aug 4 filing gave every infra founder the same public information at the same moment. The advantage now is in reading the CEO's carve-out closely, understanding that the addressable pool is ~180 and not 390, and moving on the geographies where the window is still open.

FAQ

How many Nutanix engineers are actually available from the Aug 4 cut?

The 8-K discloses ~390 total departures, but the CEO explicitly protected AI, NKP, external storage, and customer-facing sales. That leaves the addressable engineering pool concentrated in legacy HCI, AHV/Prism maintenance, non-NKP platform tooling, and support. Refolk's index suggests the NKP-adjacent engineering population company-wide is on the order of 180 people, and only a fraction of that group is exiting. Plan your outreach against a real pool in the low double digits per geography, not against 390.

Why does Bengaluru matter more than San Jose for this sourcing race?

Refolk's index shows roughly 4x more Nutanix engineers in Bengaluru than San Jose, and Bengaluru accounts for about 32% of the sampled engineering pool. India exits also move faster than US or EU exits because there is no works council process. Bengaluru engineers on platform teams skew toward AHV and Prism veterans, exactly the profile most likely to be exiting under the CEO's carve-out. US-only sourcing misses the largest and fastest-moving segment.

Does the October deadline apply to everyone?

No. The 8-K specifies that scope and timing vary by jurisdiction and are subject to works council and consultation processes. US roles close by end of October 2026, but France, Germany, and Netherlands exits typically finalize between November 2026 and January 2027. Munich, Paris, and Amsterdam are a hidden second window for anyone willing to run outreach past the US deadline.

What search terms actually find these engineers?

Search on project and product names, not marketing brands. Karbon (the pre-NKP name), AHV, Prism, Kanister, and D2iQ lineage keywords surface the real veterans. "NKP" barely appears in profile headlines even among Nutanix engineers themselves. Cross-reference against known adjacent employers - Google, Microsoft, Meta, Rubrik, and New Relic - to filter for the alumni-network overlap that predicts warm-intro conversion.