GenAI Daily - February 6, 2026: Alphabet's $180B AI Bet, OpenAI's GPT-5.3-Codex Launch, Software Selloff Accelerates
Top Story
Google Parent Doubles Down on AI with Record $180 Billion Infrastructure Bet
Alphabet announced its 2026 capital expenditures would land in a range of $175 billion-$185 billion, roughly double the $91.4 billion the company invested in 2025.
Analysts tracked by Bloomberg expected Alphabet, Google's parent company, to spend just under $120 billion.
Alphabet stock fell as much as 5% early Thursday as the tech giant's spending plans blew past forecasts, leaving investors questioning the scale and sustainability of Big Tech's investment plans to win the AI race.
Alphabet CFO Anat Ashkenazi said the higher 2026 spending would go toward AI computing infrastructure as the company looks to develop frontier AI models and meet demand for its Cloud and Services segments.
The release of Google's Gemini 3 AI model - which outperformed competing models on benchmark tests and prompted rival OpenAI to declare a "code red" - along with RBC analyst noting that the momentum in the Gemini app and spike in fourth quarter Google Cloud revenue were "plenty good as proof points which warrant the higher spend."
Fourth quarter revenue climbed 18% to $113.8 billion from the year-ago period, ahead of the $111.4 billion expected by analysts, with earnings per share rising to $2.82 from $2.15 in the previous year.
Why it matters: Alphabet's infrastructure spending signals the arms race for AI compute has no ceiling, potentially reshaping enterprise vendor negotiations as cloud costs transfer to customers.
Key Developments
OpenAI Ships GPT-5.3-Codex with 25% Speed Boost - UPDATE
OpenAI's new coding flagship is live: GPT-5.3-Codex launched on February 5, 2026 across all Codex surfaces (app, CLI, IDE extension, web) for paid ChatGPT plans, with API access announced for the coming weeks.
GPT-5.3-Codex combines the frontier coding performance of GPT-5.2-Codex with the reasoning and professional knowledge capabilities of GPT-5.2 into a single model that is also 25% faster.
The team used early versions of the model to debug its training, manage its deployment, and diagnose test results and evaluations, with the team reporting they were "blown away" by how much Codex was able to accelerate its own development.
The model shows a large jump on terminal and computer-use tasks with 77.3% on Terminal-Bench 2.0 and 64.7% on OSWorld-Verified, while SWE-Bench Pro leadership remains incremental at 56.8% versus 56.4% for GPT-5.2-Codex.
(OpenAI)
Impact: Self-improving AI development cycles could accelerate model releases and challenge traditional software development timelines.
Anthropic Releases Claude Opus 4.6 with Agent Teams
Anthropic released the latest version of Opus - its most advanced model particularly important for Claude Code, with Opus 4.5 only released last November, and with 4.6, the company has sought to broaden its model's capabilities and appeal.
The most notable addition is "agent teams" - teams of agents that can split larger tasks into segmented jobs, allowing multiple agents to coordinate in parallel rather than working sequentially.
Opus 4.6 is the first model in the Opus family to feature a one-million-token context window and the first Anthropic model to use adaptive thinking.
The standout performance is its 68.8% score on the ARC AGI 2 benchmark for solving problems that are easy for humans and very hard for AI systems, compared to Opus 4.5's 37.6%, Gemini 3 Pro's 45.1%, and GPT-5.2's 54.2%.
The new version integrates Claude directly into PowerPoint as an accessible side panel, allowing presentations to be crafted within PowerPoint with direct help from Claude.
Impact: Parallel agent coordination could reshape enterprise workflow automation, challenging traditional sequential task management systems.
Software Stocks Plunge as AI Disruption Fears Escalate
After a broad selloff on Tuesday that saw the S&P 500 software and services index fall nearly 4%, the sector slipped another 0.73% on Wednesday, notching the sixth straight session of losses and wiping out about $830 billion in market value since January 28.
The WisdomTree Cloud Computing Fund has plummeted about 20% so far in 2026, including a 6.5% drop this week, as AI concerns punish the sector.
HubSpot has fallen 39% this year following a 42% slump in 2025, while Figma has plunged 40% this year, Atlassian is down 35%, and Shopify has dropped 29%.
The anxiety was triggered Tuesday after AI startup Anthropic released a productivity tool for in-house lawyers, sending legal software and publishing firms tumbling, with London Stock Exchange Group falling 13%, Thomson Reuters plunging 16%, CS Disco sinking 12%, and Legalzoom.com plummeting 20%.
The tool underscored the push by LLMs into the "application layer," where these firms are increasingly muscling into lucrative enterprise businesses for revenue they need to fund massive investments.
Impact: The selloff signals investors are repricing software companies for an AI-disrupted future, potentially creating acquisition opportunities as valuations compress.
Okta Flags Authorization Gap in AI Agent Security
AI agents in shared workspaces expose sensitive data due to an authorization gap, with Okta highlighting how fine-grained authorization can fix this by scoping permissions before data retrieval.
As AI agents use protocols like Model Context Protocol (MCP) and Agent2Agent (A2A) to connect to other systems, human-centric access models have already reached their limits, with these agents moving between environments, accessing data, and initiating workflows in ways that are difficult to track, let alone govern.
With 91% of organizations deploying AI agents but only 10% having well-developed governance strategies, the risk isn't just emerging - it's already here.
Beginning August 2, 2026, enforcement of Article 14 of the EU AI Act will require organizations to prove that every AI-driven action was authorized at the time it occurred, not just when credentials were issued, with violations costing up to €35 million ($38 million) or 7% of global revenue.
(Okta)
Impact: The authorization gap represents a critical enterprise security exposure as AI agents proliferate without proper identity controls ahead of regulatory enforcement.
Funding & Deals
- Cerebras Systems - $1B Late-Stage:
The California-based chipmaker designing massive wafer-scale processors for AI model training raised $1 billion in late-stage financing, valuing the company at $23.1 billion. Tiger Global led the round, with Benchmark, Coatue, and 1789 Capital joining. This is Cerebras's second billion-dollar round in six months.
- Waabi - $750M Series C:
The Toronto-based startup pioneering "Physical AI" for self-driving trucks and robotaxis secured a massive $750 million Series C, the largest-ever tech fundraising in Canadian history. The funding comes with a strategic partnership with Uber to exclusively power 25,000+ Uber robotaxis over time.
Led by Khosla Ventures (Tech Startups)
- Bedrock Robotics - $270M Series B:
The company builds retrofit kits that turn heavy construction machinery like excavators into autonomous, coordinated fleets. The Series B round will expand beyond single-machine deployments and target fully operatorless excavator operations in 2026. Co-led by CapitalG (Google's growth fund) and the Valor Atreides AI Fund.
Related reading: Check out this week's Deep Insights analysis for strategic context on these developments.
