Top Stories
Snap Cuts 1,000 Jobs, Says AI Now Writes 65% of Its Code
Snap is eliminating 16% of its full-time workforce — roughly 1,000 people — along with 300 open positions. CEO Evan Spiegel framed the cuts around AI, telling employees that “rapid advancements in artificial intelligence” allow smaller teams to match the output of larger ones. The company says AI now generates more than 65% of its new code.
Wall Street rewarded the move. Snap shares jumped about 7% on the announcement, even though the stock is still down more than 30% in 2026. Activist investor Irenic Capital Management, which holds a 2.5% stake, had been pushing for cost reductions. Affected U.S. employees will receive four months of severance, healthcare coverage, and equity vesting.
The 65% code generation figure is worth examining. It is a measure of lines produced, not of shipping quality, security, or architectural decisions. When a company uses AI-generated code volumes as justification for cutting the humans who review and maintain that code, the math stops working somewhere. Snap expects to save $500 million annually from the reduction — and investors are pricing in the savings without pricing in the risk.
Sources: TechCrunch, CNBC, TechRepublic
PwC Study: 20% of Companies Capture 74% of AI’s Economic Value
PwC’s 2026 AI Performance Study, surveying 1,217 senior executives across 25 sectors, found that the top 20% of companies generate 7.2 times more AI-driven revenue and efficiency gains than the average competitor. The other 80% are largely stuck in pilot mode or seeing marginal returns.
The distinguishing factor is not spending. Leaders are nearly twice as likely to use AI in advanced autonomous workflows and are increasing the number of decisions made without human intervention at 2.8 times the rate of their peers. They also invest more heavily in data foundations and governance — the unsexy infrastructure that makes AI outputs reliable enough to act on.
The study’s most telling number: companies pursuing “industry convergence” opportunities — using AI to enter adjacent markets — outperform those focused purely on internal efficiency. AI that replaces existing processes saves money. AI that opens new revenue lines grows the business. Most companies are only doing the first.
Sources: PwC
NVIDIA Releases Ising: Open AI Models for Quantum Computing
NVIDIA launched Ising, the first family of open AI models designed specifically for quantum processor development. Released under NVIDIA’s Open Model License on GitHub and Hugging Face, Ising targets the two biggest engineering bottlenecks in quantum computing: calibration and error correction.
Ising Calibration is a 35-billion-parameter vision-language model that analyzes quantum calibration experiment output and operates in agentic workflows with minimal human oversight. NVIDIA says it can compress calibration work from days to hours. Ising Decoding consists of two models — one tuned for speed, one for accuracy — delivering quantum error-correction decoding up to 2.5x faster and 3x more accurate than traditional methods.
The practical significance: quantum computing’s biggest barrier is not physics concepts but engineering execution. Calibrating a quantum processor is tedious, error-prone, and requires deep domain expertise. If Ising delivers on its claims, it lowers the expertise threshold for building functional quantum hardware — which matters more for the field’s timeline than any single qubit milestone.
Sources: NVIDIA Newsroom, NVIDIA Developer Blog
Quick Hits
- Novo Nordisk partnered with OpenAI to deploy AI across drug discovery, clinical trials, manufacturing, and supply chains, with pilot programs launching immediately and full integration planned by end of 2026. The deal includes strict data governance provisions but no disclosed financial terms. CNBC, GlobeNewsWire
- Disney laid off roughly 1,000 employees in new CEO Josh D’Amaro’s first major move, cutting across Marvel (7% of staff), ESPN, marketing, and corporate functions. While framed as operational streamlining rather than AI replacement, D’Amaro cited the need for “a more agile and technologically-enabled workforce.” Deadline, Popverse
- The GUARDRAILS Act was introduced in both chambers of Congress to block the White House’s proposed moratorium on state-level AI regulation. The bill would repeal the Trump administration’s AI policy framework that sought federal preemption over state AI laws. Rep. Beyer
- Colorado’s AI Act is on track for June 30, 2026 enforcement, requiring companies to demonstrate reasonable care against algorithmic discrimination and implement risk management frameworks — making it the first comprehensive state AI law to take effect. Gunderson Dettmer
Worth Watching
The Snap layoffs set a template that other companies will follow: cite AI code generation metrics, cut headcount, and let the stock price validate the decision. The 65% figure will be repeated in earnings calls across the industry. What will not be tracked is how many of those AI-generated lines of code create maintenance debt, security vulnerabilities, or architectural drift that the remaining engineers inherit.
The PwC 74/20 split has policy implications beyond business strategy. If AI’s economic gains concentrate this sharply, the technology accelerates market consolidation rather than enabling broad-based competition. Regulators focused exclusively on AI safety are missing the antitrust dimension. Colorado’s AI Act, taking effect in ten weeks, is the first state attempt to address this — but its focus on algorithmic discrimination does not touch economic concentration.