By mid-2026 artificial intelligence is no longer a bet on the future: it is the engine steering tech markets. But behind the record numbers lies a subtler picture — hyper-concentrated capital, soaring memory prices and shifting rules. Let's read it in order, with figures checked at the source.

The backdrop: record capital, ever more concentrated

The first half of 2026 rewrote the venture-capital record books. According to Crunchbase, global startup investment hit a record $510 billion in the first six months — more than was raised in all of 2025. In the United States, venture capital deployed $412.7 billion, of which 86% — roughly $356 billion — went to AI companies.

Yet the figure that matters most is not size but concentration. Two companies alone, OpenAI and Anthropic, absorbed around 43% of all global venture capital in the half. Growth did not come from a broadening market — deal count barely moved — but from a few giant mega-rounds. That distinction is decisive for investors: a rally driven by a handful of names is, by definition, more fragile than a broad-based one.

The engine: semiconductors and the memory supercycle

If AI is the fuel, semiconductors are the engine. Gartner expects global industry revenue to exceed $1.3 trillion in 2026, growing 64% — the fastest in two decades — with AI-specific chips making up about 30% of the total. The push comes from hyperscalers, whose AI-infrastructure spending is set to rise more than 50% this year.

That demand has triggered what analysts call "memflation": an unprecedented spike in memory prices. Gartner estimates 2026 average annual increases of 125% for DRAM and 234% for NAND flash. Quarterly data confirm the surge — DRAM contracts jumped nearly 60% in the second quarter — though a slowdown is emerging from the third quarter, with rises easing to low double digits. Normalisation, analysts warn, is unlikely before 2027-2028.

The semiconductor ETF chart above (VanEck Semiconductor, ticker SMH) captures how equity markets have priced this cycle. Read it for what it is: an illustrative snapshot of the sector's trend, not trading advice on any single security.

Maturity: from hype to operations

Compared with the years of pure experimentation, 2026 marks AI's move from pilot projects to integration into companies' core processes. The dominant theme is agentic AI: systems that carry out tasks and coordinate decisions with limited human oversight. Finance, healthcare and marketing lead adoption in search of efficiency and productivity. But the technology is outrunning organisations' ability to absorb it: scarce skills and fragmented data remain the main brakes.

The risks: bubble, energy and changing rules

Concentration and bubble fears

The question hanging over Wall Street is always the same: how much of this value is real? The gap between the vast sums invested in AI infrastructure and still relatively modest actual revenue fuels bubble fears. Unlike 2000, much of the spending is backed by companies with solid profits and ample cash; but rising leverage and concentration in a few names are factors a prudent investor cannot ignore.

The energy bill

AI data centres have an appetite for electricity that is becoming a strategic issue, not just an environmental one. The most-cited estimates point to consumption approaching a meaningful share of global electricity by the end of the decade: a physical constraint that can slow growth and push up costs.

A moving regulatory picture

Here lies one of the most important — and often outdated — developments. The EU AI Act does not enter into force for high-risk systems in August 2026, as was reported until recently. Under the "Digital Omnibus" simplification package, given final approval by the EU Council on 29 June 2026, the obligations have been postponed:

• Stand-alone high-risk systems (Annex III): compliance required from 2 December 2027.
• AI embedded in regulated products (Annex I): from 2 August 2028.
• Penalties: up to €15 million or 3% of global annual turnover.

For the sector this delay is operational breathing room; for investors it is one more variable to track, as it reshapes compliance timing in Europe.

The bottom line: reading 2026

2026 is the year AI stops being a story and becomes a reality check. The fundamentals are real — revenue, adoption, infrastructure — but the price of exposure is high and concentrated. Discipline, more than enthusiasm, will likely separate those who navigate this cycle from those who get burned.

Disclaimer: this article is for information purposes only and does not constitute financial advice. Any investment decision should be assessed against your own circumstances and, if needed, with a qualified professional.