Zymbos Intelligence — Issue 003

Welcome to Issue 003 of Zymbos Intelligence.

This week, something shifted. Not in the technology — in the money behind it.

The UK government committed £2.5 billion to AI and quantum. NVIDIA announced $1 trillion in chip orders through 2027 — double what it projected twelve months ago. Meta is cutting over 15,000 people to fund its AI infrastructure build. OpenAI is preparing for a public listing that could value it at $1 trillion. And the UK's long-awaited AI and copyright report landed today, raising the question of who gets to profit from the intellectual capital used to build these systems in the first place.

Every story this week is a capital allocation story. The organisations, governments, and investors that understand this are moving. The ones still treating AI as a pilot programme are falling behind — not on technology, but on something harder to close: capital position.

This week's Tool on Trial is Perplexity Pro — the research engine that may already be the most practical AI investment a professional can make at £15 a month. And the Prompt Pocket gives you a structured audit of your own AI exposure, so you can answer the capital question for your own role before someone else answers it for you.

Let's get into it.

Intelligence Briefing

Five stories that matter this week

01  |  Policy & Investment  |  UK

UK Chancellor Pledges £2.5 Billion to Win the AI Race

In a major policy speech this week, Chancellor Rachel Reeves committed a record £2.5 billion to AI and quantum technologies, pledging the UK will achieve the fastest AI adoption in the G7. The package includes a £500 million Sovereign AI Fund to give British companies access to compute and funding, and £2 billion to upgrade the UK's quantum capabilities, including a programme to procure commercial-scale quantum computers. The move signals a decisive government push to anchor strategic AI industries in the UK and positions Britain as a serious competitor in the global race for AI infrastructure dominance.

The Sovereign AI Fund is the right instinct. Compute access is the bottleneck for British AI businesses — not ambition. Whether the execution matches the announcement is the question to watch.

Read the full story (GOV.UK)

02  |  Infrastructure  |  US / Global

NVIDIA Sees $1 Trillion in Orders for Next-Generation AI Systems Through 2027

At NVIDIA's GTC 2026 conference this week, CEO Jensen Huang announced the company anticipates $1 trillion in purchase orders for its Blackwell and upcoming Vera Rubin chip systems through 2027 — doubling the $500 billion projection made at GTC 2025. Key unveilings included the Vera Rubin architecture, promising a 10x performance-per-watt improvement, and the Groq 3 LPU, NVIDIA's first inference chip from its recent $20 billion Groq acquisition. Huang also introduced NemoClaw, an enterprise-ready open source platform for building autonomous AI agents, and confirmed Uber will launch an NVIDIA-powered autonomous vehicle fleet across 28 cities by 2028.

A trillion dollars in orders through 2027. The infrastructure race is not slowing — it is accelerating. Any organisation still treating AI as a pilot programme is now operating in a different economic reality to its competitors.

Read the full story (CNBC)

03  |  Workforce & Enterprise  |  US

Meta Planning Sweeping Layoffs Affecting Over 15,000 Employees as AI Costs Mount

Meta is reportedly planning to cut more than 20% of its workforce — over 15,000 employees — to offset the cost of its massive AI infrastructure investment, including a planned $135 billion data centre spend in 2026 alone. CEO Mark Zuckerberg has stated that tasks once requiring large teams can now be completed by a single person using AI. The planned cuts follow similar restructuring at Atlassian, which announced 1,600 redundancies on 12 March, and Block, which cut 4,000 employees in February. Global tech layoffs in 2026 had already surpassed 45,000 by early March, with AI cited as a structural driver across the sector.

This is not a cost-cutting story dressed up as an AI story. It is the first wave of a structural labour shift that most organisations are not yet prepared to discuss internally, let alone plan for.

Read the full story (Reuters)

04  |  Governance & Legal  |  UK

UK Government Publishes AI and Copyright Report — and Leaves the Hard Questions Unanswered

Today marks the legal deadline under the Data (Use and Access) Act 2025 for the UK government to publish its report on AI and copyright — assessing the policy options on transparency, licensing, and the use of copyrighted material to train AI models. The Lords Communications and Digital Committee published its own inquiry findings on 6 March, discouraging the government from introducing a new text and data mining exception and calling for statutory transparency requirements on AI training data. Technology Secretary Liz Kendall has signalled the government has not yet reached a decision on its preferred approach.

The UK is doing the right thing by not rushing. A bad copyright framework for AI would damage both the creative sector and the AI sector simultaneously. But the window for getting this right is narrowing as the US and EU move ahead.

Read the full story (techUK)

05  |  Business & Finance  |  US

OpenAI Prepares for IPO by End of 2026 — and Repositions ChatGPT as a Productivity Tool

OpenAI is preparing to go public by the end of 2026, with the company having selected law firms to lead preparations for a listing that could value it at up to $1 trillion. Alongside the IPO preparations, the company is internally repositioning ChatGPT as a productivity tool — a deliberate shift intended to strengthen its enterprise appeal. OpenAI crossed $25 billion in annualised revenue at the end of February 2026, a 17% increase in two months, though the company is not expected to reach profitability until 2030. Rival Anthropic, approaching $19 billion in annualised revenue, is targeting breakeven two years earlier in 2028.

The IPO will be the first major test of whether public markets can accurately price frontier AI infrastructure. The gap between OpenAI's revenue growth and its profitability timeline is the story most investors are not yet asking enough questions about.

Read the full story (CNBC)

Deep Intelligence

The Efficiency Dividend Has Arrived. Most Organisations Have No Plan for It.

This week, three of the world's most prominent technology companies announced or confirmed large-scale workforce reductions. Meta is planning to cut more than 15,000 people. Atlassian cut 1,600. Block cut 4,000 in February. In each case, the stated rationale was the same: AI is changing the mix of skills we need and the number of roles required to do the work.

At the same time, Jensen Huang stood in front of a packed arena in San Jose and announced $1 trillion in orders for the infrastructure that will power the next generation of AI systems. The two stories are not separate. They are the same story told from opposite ends of the economic chain.

The efficiency dividend — the productivity gain that AI proponents have been promising for three years — is arriving. The question organisations are not yet asking loudly enough is: what happens to the people on the wrong side of it?

The pattern is becoming impossible to ignore

What is significant about this week is not that Meta is cutting jobs. Large companies restructure constantly. What is significant is the reasoning being given openly and on the record by senior leaders.

Mark Zuckerberg has said tasks once requiring large teams can now be completed by a single person. Mike Cannon-Brookes at Atlassian acknowledged that AI changes the number of roles required in certain areas. Jack Dorsey at Block declared a move to an intelligence-native operating model and cut nearly half his workforce.

These are not cautious, hedged statements. They are leaders publicly articulating a strategic shift in how they view the relationship between their business and their headcount. That shift is now moving from Silicon Valley into every sector.

The organisations that handle this transition well will retain the trust of their remaining workforce at exactly the moment when human judgment, institutional knowledge, and cultural continuity matter most.

What leaders should be doing now

Map the efficiency dividend before it maps you. Identify which roles in your organisation are most exposed to AI-driven task automation in the next 12 to 24 months. Do this analysis before external pressure forces the conversation.

Separate task automation from role elimination. Most roles will change before they disappear. The organisations getting this right are redesigning workflows around augmented capability, not simply reducing headcount.

Build the internal narrative now. How your organisation talks about AI and its workforce implications internally will determine whether you retain the talent and trust you need to execute on the strategy. Silence is not neutral — it reads as evasion.

The efficiency dividend is real. The organisations that capture it while maintaining the confidence of their people will be in a structurally stronger position in three years than those that treat this purely as a cost reduction opportunity.

Tool on Trial

This Week

Perplexity Pro

AI Research and Intelligence  ·  Zymbos Score: 7.9

If your primary AI workflow involves finding, synthesising, and verifying information at speed, Perplexity Pro is the most capable tool available for that specific job in 2026. It is not a general-purpose AI assistant. It is a research engine — and within that lane, nothing currently matches it.

The core proposition is straightforward. Where a standard AI model generates an answer from training data, Perplexity retrieves live information from across the web, synthesises it from multiple sources simultaneously, and presents the result with clickable citations so you can verify every claim. For professionals who need to stay current on fast-moving topics — AI developments, regulatory changes, competitor activity, market signals — this is a meaningfully different capability.

The Pro plan adds access to premium AI models within a single interface, including Claude, GPT-5, and Gemini, switchable per query. The February 2026 addition of Model Council allows you to run the same query through multiple models simultaneously and compare responses side by side. Deep Research mode, available on Pro at 20 queries per day, conducts multi-step investigations across dozens of sources and returns a structured research brief.

What it does well: Real-time web retrieval with source citations. Multi-model access under one subscription. Handling complex, layered research queries. Speed and ease of adoption — G2's Winter 2026 data puts ease of use at 95% and ease of setup at 97%.

What it does less well: Synthesis and original analysis. Creative, long-form, or generative tasks. Citation accuracy at the individual claim level — a Columbia Journalism Review benchmark found a 37% citation error rate on the Pro plan, the best result of any tool tested but still a meaningful limitation. Verification remains your responsibility, not the tool's.

The right use case: Daily intelligence gathering, competitive research, regulatory monitoring, and pre-meeting briefing. Not first-draft content creation, strategic analysis requiring original synthesis, or coding.

Best for: Professionals, analysts, consultants, and leaders who need to stay informed across multiple domains without spending hours in browser tabs.

Pricing

Individual plans: Free ($0). Pro — £15/month ($20). Education Pro — £7.50/month ($10) with verified student or faculty status. Max — £150/month ($200) or £1,499/year ($2,000), annual billing via web only.

Enterprise plans: Enterprise Pro — £30/seat/month ($40) or £299/seat/year ($400). Enterprise Max — £243/seat/month ($325) or £2,433/seat/year ($3,250). Educational institutions and nonprofits: Enterprise Pro at £22/seat/month ($30) or £224/seat/year ($300).

Sterling figures based on the mid-market rate of $1 = £0.75 as of 18 March 2026. Verify current pricing at perplexity.ai.

Verdict: Recommended — with verification. Perplexity Pro compounds over time. The more consistently you use it for structured research rather than casual queries, the more value it delivers. But treat every citation as a starting point, not a conclusion.

Try Perplexity Pro

The Prompt Pocket

Map Your AI Exposure Before Someone Else Does It For You

Strategy  ·  Workforce  ·  Leadership

Use this prompt to identify which parts of your role or organisation are most exposed to AI-driven task automation — and which are most likely to compound in value. Works best with Claude or Perplexity Pro with web search enabled.

McGann's Take

John McGann

Founder, Zymbos AI  ·  Programme Director & AI Innovator  ·  London

This week, AI stopped being a technology decision and became a capital decision.

That is the thread connecting every story in this issue. The UK government commits £2.5 billion. NVIDIA announces $1 trillion in chip orders through 2027 — double what it projected twelve months ago. Meta cuts 15,000 people specifically to redirect that spend into AI infrastructure. OpenAI prepares for a public listing that could value it at $1 trillion. And the UK copyright report — which looks at first glance like a legal and creative industries story — is fundamentally a question of who gets to profit from the intellectual capital used to build these systems.

Every story this week is a capital allocation story.

For three years, organisations have treated AI adoption as a strategic choice with an open timeline. This week closes that window. When governments, infrastructure providers, enterprise software companies, and the world's most valuable AI startup all make capital commitments of this magnitude in the same week, the cost of inaction stops being theoretical and starts appearing on a balance sheet.

Organisations that have not yet moved AI from pilot to infrastructure are not behind on technology. They are behind on capital allocation. And that is a significantly harder problem to fix. Technology gaps close quickly. Capital gaps compound in the wrong direction.

The question I would be asking in every board meeting and leadership team right now is not "are we using AI?" It is "where are we allocating capital relative to what this week tells us about the pace and scale of what is coming?" The organisations that answer that question honestly — and act on the answer — will be in a structurally different position in three years to those that do not.

The race is not coming. The capital is already deployed.