FAANG digest, Jun 29-Jul 6: layoffs lead

Microsoft is the week’s biggest candidate-risk signal. The company is expected to cut about 5,000-5,700 jobs across sales, consulting, and Xbox during the week of July 7, below 2.5% of a global workforce reported around 220,000-228,000. 12 That matters even for SWE candidates because the same week also brought SAP freezing non-AI hiring, Google showing more process anomalies, and Meta producing its first fresh SWE interview signal in weeks. 345

Coverage runs from Jun 29 at 9:45 a.m. through Jul 6 at 9:00 a.m. Eastern. The practical read: candidates should still prepare for active loops, but they should price timing risk more aggressively. Google candidates need clearer team-match milestones. Meta candidates finally have a current AI-enabled coding example. Microsoft, SAP, ServiceNow, and Salesforce candidates should ask whether the role is funded, AI-aligned, and tied to a team that survived the current reset.

The highest-engagement post this week was not a single-company question bank. It was a 7-YOE ex-2-FAANG engineer’s post after failed Anthropic and Databricks interviews, with 133 upvotes and 77 comments. The claim was blunt: for roles above $500K, close answers are no longer enough, and code has to run correctly. 19

Levels.fyi medians are still the cleanest cross-company benchmark, but the offer posts explain why candidates are anchoring differently for AI labs and Amazon AGI. OpenAI’s Levels.fyi L5 SWE median is $869K, while a self-reported OpenAI Senior SWE offer reached about $1.012M first-year TC for 4 YOE. 1621 Amazon’s L7 median stayed at $751K, but an Amazon AGI L7 self-report reached about $1.153M first-year TC because the Y1 signing bonus alone was $780K. 2223

The self-reported offer set adds three negotiation notes. Amazon AGI L7 in San Francisco reported $300K base, $780K Y1 signing, $610K Y2 signing, and $1.46M equity over 4 years for about $1.153M first-year TC. 23 OpenAI Senior SWE in San Francisco reported $325K base, $100K signing, and $2.35M equity over 4 years for about $1.012M first-year TC. 16 xAI Staff SWE in Seattle reported $200K base and $1.6M equity over 4 years for $600K first-year TC, and the candidate declined. 33

Microsoft is the timing problem. Reuters/AOL reported the company planned to cut under 2.5% of its workforce, while GeekWire confirmed the cuts would span Xbox, sales, and consulting. 12 The Microsoft signal also comes after a spring voluntary retirement program in which roughly one-third of about 8,750 eligible U.S. employees accepted buyouts. 2

SAP is the clearer role-filtering signal. The company’s internal memo said new hiring would focus only on selected profiles, mainly core AI roles, and CEO Christian Klein told The New York Times he was not sure whether someone at SAP would still code software in two or three years. 3 For candidates, that makes the first recruiter question simple: is this role inside the AI hiring lane or outside it?

Cisco shows the same reallocation pattern from a different angle. California WARN notices showed 471 Bay Area cuts effective July 13, with 56 software engineers and 39 software engineering technical leaders among the affected roles, while Cisco had reported record Q3 FY2026 revenue of $15.841B and GAAP net income of $3.373B. 34 That is not a classic demand-collapse signal; it is a role mix signal.

ServiceNow and Salesforce remain contradiction cases. ServiceNow’s rumored July 6 third wave was not confirmed at the check time, but California WARN data showed 117 permanent cuts effective Aug. 17 and Blind discussion pointed to July 29 as another date employees were watching. 18 Salesforce, meanwhile, announced 1,000 AI/Agentforce graduate and intern roles while also carrying June cuts affecting 86 California employees and reported Agentforce-team reductions. 17

Google’s risk is procedural rather than layoff-led. One candidate said a recruiter could not disclose onsite outcome until the candidate first matched with a team and then went to Hiring Committee. 4 Another L5 non-SWE candidate reported an added skip-skip-level chat after onsite, with notes taken and extra days needed for the decision. 6 The safest interpretation is not that Google has a universal new process; it is that candidates need written expectations for HC, team match, extra rounds, and expiry timing.

If your Google loop is already complete, ask the recruiter to separate four milestones: onsite feedback, HC packet timing, team-match order, and whether additional interviews can still be added. The week had team-match-before-HC, skip-skip follow-up, and long HC wait reports, so vague "still in process" language is not enough for planning other offers. 468

If your Meta loop is mid-level SWE, practice with AI in the loop and without it. The reported onsite allowed or involved AI in one coding round, but also included regular coding and system design. 5 The practical bar is not "can the model help?" It is whether you can define the problem, validate edge cases, and explain why the output is correct.

If your Anthropic loop is coming up, do not treat the culture round as a soft close. The Staff SWE Infra report’s hardest questions were not standard conflict stories; they asked about disliked work, persuasion around disliked work, and morally incorrect behavior. 13 Prepare examples that a non-engineer can understand without losing technical substance.

If you are negotiating an AI-lab or Amazon AGI offer, separate three numbers: annualized median TC, first-year cash, and equity liquidity. Amazon AGI’s reported first-year package is cash-heavy because of a $780K Y1 signing bonus, while OpenAI’s reported Senior SWE package is equity-heavy because $588K of first-year value comes from equity. 1623

If you are interviewing at Microsoft, SAP, Salesforce, or ServiceNow, ask whether the role is funded, which budget owns it, and whether the hiring plan changed after the latest AI or layoff review. That question is not defensive theater this week. It directly follows from Microsoft’s expected July cuts, SAP’s non-AI hiring freeze, Salesforce’s AI-hiring-plus-cuts split, and ServiceNow’s unresolved layoff timing. 131718

The edge this week is not memorizing one more question bank. It is matching your prep and negotiation behavior to the company’s current failure mode: Google process drift, Meta AI-enabled evaluation, Anthropic culture-plus-systems selectivity, Amazon OA ambiguity, and enterprise-software headcount resets.