The Source Signals
Tech news is everywhere. Intelligence isn't.
LEAP 2026: HUMAIN will replace Windows with its own oS on the Horizon Ultra in 2027
The Saudi company launched an AI PC with Qualcomm at LEAP 2026, running Microsoft software first and its own operating environment from next year.
LEAP 2026: Riyadh became the first Lenovo site to build the company's full product range
The plant behind Lenovo's first Saudi-made laptop will produce PCs, desktops, Motorola smartphones and servers, a combination no other facility in its network of more than 30 sites handles.
LEAP 2026: Four rival chipmakers agreed to build Saudi Arabia's AI stack at once
Across four days in Riyadh, HUMAIN secured commitments from AMD, NVIDIA, Qualcomm and Cisco, and placed its Arabic language model in both Amazon's and Microsoft's catalogues. Assembling all of them in one week is the most instructive thing that happened at LEAP 2026.
SandboxAQ turns team chat into the shared workspace for AI agents with Switch
Switch connects agents built on Claude Code, Google ADK, LangChain and OpenAI into Slack, Microsoft Teams and Discord at no cost, under a licence that reserves commercial rights for SandboxAQ.
Mistral and HUMAIN sign sovereign AI deal worth hundreds of millions of euros
The Saudi collaboration covers infrastructure, Arabic-language models and regulated industries, and comes 13 days after Mistral locked five European industrial groups into multi-year compute commitments.
The AI dividend enterprise IT keeps promising has not turned up
New global research across 844 IT professionals records real productivity gains from AI in service management alongside a workload that has stayed flat or grown for seven in 10 teams. The measurement regime, not the model, explains most of the difference.
Agent sprawl drags enterprise workflow integration down 14 points
The share of organisations running AI-integrated workflows across business functions fell from 30% to 16% over the past year even as AI budgets grew 110%, according to ServiceNow’s third Enterprise AI Maturity Index.
AI is blamed for a quarter of all job cuts this year. The labour data cannot find the effect anywhere
Over-hiring during the cheap-money years and ordinary margin pressure are being consolidated into a single flattering story about operating at the frontier, and almost nobody outside the company can check it.
A sovereign AI system is only as sovereign as the person using it
Two years of architecture work has put data inside national borders, and the control deciding whether those borders hold sits with an employee and their training.
NVIDIA turns the AI factory into an asset class with $500 billion financing push
Partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR are designed to mobilise more than $500 Billion of third-party capital, shifting the AI buildout from corporate balance sheets towards the institutions that finance power grids and ports
Alphabet is now funding the startup built by the researchers it lost
Jeff Dean has left Google after 27 years to co-found Discovery Loop with three colleagues, and Alphabet has taken a founding investment alongside a cloud partnership, on the same day Demis Hassabis gave up operational control of Google DeepMind.
ServiceNow bets on security consolidation as AI compresses response times
ServiceNow has brought exposure management, identity, cyber-physical security, incident response and compliance into one governed system, wagering that enterprises will trade specialist tool depth for auditable autonomy.
Why 1,293 frontier AI employees, Dario Amodei included, asked Washington for tools to pace AI development
The July statement asks the US government for the technical and governance tools to pace automated AI development, because nobody will slow down while rivals cannot be seen doing the same. Dario Amodei signed alongside four Anthropic co-founders.
Jensen Huang breaks his X silence to make open-weight AI a test of American leadership
The Nvidia chief executive used the first post of his life on the platform to circulate a coalition letter urging Washington to keep downloadable AI models legal, arguing that the country's lead depends on diffusing open models across the economy.
“Scientists have felt left out of the AI craze. That’s the chasm we’re closing”, Bill Fitzgerald, SandboxAQ
SandboxAQ’s Bill Fitzgerald on bringing Large Quantitative Models to Google Cloud’s Marketplace, the reasoning behind pairing with rival frontier LLMs, and why materials discovery reaches the market before drug discovery.
The middle layer, not the model, now decides whether enterprise AI works
As foundation models converge toward parity, Zoho and TERN Group argue that the orchestration, memory and guardrail layer built around them is the real source of durable advantage.
The Token Cost Crunch: Why Falling AI Prices Are Producing Rising AI Bills
Pilots run cheap, but production does not. Close to 73% of organisations exceeded their AI cost projections in 2026 after committing to architectures they costed at a fraction of real volume.
Demis Hassabis says the age of thinking machines is almost here, and the field cannot agree whether he is right
Google DeepMind's chief executive has published his fullest statement yet on why he believes artificial general intelligence is only a few years away and what it would change. The claim carries the weight of a Nobel laureate at the head of a leading lab, and it remains one of the most contested propositions in the field.
Why Poor Data Quality Makes Most Enterprise AI Fail
Two decades of cleaning data have left it in worse shape than ever, and AI has made the failure impossible to hide. Industry leaders explain what actually needs to change.
One AI system, many rulebooks: how data residency is rewriting global vendor strategy
Multinationals are learning that a single AI architecture cannot cross borders intact, as data residency mandates in the EU, India and the Gulf split vendor strategy and shift advantage to providers built for sovereignty from the start.