AI Costs Surge, Microsoft Copilot Repriced & Meta's 6,000 Cuts
Failed to add items
Sorry, we are unable to add the item because your shopping cart is already at capacity.
Add to basket failed.
Please try again later
Add to wishlist failed.
Please try again later
Remove from wishlist failed.
Please try again later
Adding to library failed
Please try again
Follow podcast failed
Unfollow podcast failed
AI Costs Surge, Microsoft Copilot Repriced & Meta's 6,000 Cuts
-
Narrated by:
-
By:
(00:00:36) Hardware Relief Is Years Away
(00:01:12) Infrastructure Bets Signal Long Squeeze
(00:01:42) Microsoft Copilot Pricing Redesign
(00:02:08) AI Workforce Cuts Accelerate
(00:02:40) Regulation Steps Back
AI pricing just shifted under every business running modern tools. OpenAI has doubled the price of GPT-5.5, Google's latest Gemini Flash is three to six times more expensive than its predecessor, and the cause isn't temporary — agentic AI burns through compute at a rate that has broken the old economics. New inference-optimised hardware from Nvidia, AMD, and Intel won't reach widespread deployment until early-to-mid 2027, leaving model providers with real pricing power and no competitive pressure for at least 12 to 18 months.
The venture capital signal backs that up. Modal Labs just raised $355 million at a $4.65 billion valuation after growing annualised revenue from $60 million to $300 million in months — investors are betting on a multi-year compute squeeze, not a short-term blip.
Microsoft has responded to the same pressure by moving GitHub Copilot off per-seat pricing entirely. Usage-based models shift token-budget risk back to customers, and more platforms are likely to follow. If you're negotiating enterprise AI contracts today, the structure you lock in matters.
On workforce, the pattern has solidified. Meta is cutting around 6,000 roles, Cloudflare 1,100, and Cisco 4,000 — all citing AI efficiency gains. AI-driven headcount reduction is no longer a boardroom debate; it's being approved and announced.
Meanwhile, regulation is stepping back. The White House delayed a planned AI executive order requiring security evaluations of frontier models, citing competitive concerns. New Zealand is taking a similar stance. Governance arriving after deployment is becoming the default pattern.
The two watchpoints for the next cycle: will hardware costs actually fall on schedule, and will the workforce cuts deliver the claimed efficiency gains — or create rehiring pressure in 12 months?
This episode includes AI-generated content.
adbl_web_anon_alc_button_suppression_c
No reviews yet