Grok and Claude are winning the AI race

Five charts to start the day

Good morning – here are your five chart for the day.

I have to admit, I was a bit disappointed with DeepSeek. When it was launched I heard wonderful things about it from the tech community. It seems incredible that it could compete with the likes of ChatGPT, despite using a fraction of the resources and energy to run. Alas, it was too good to be true.

In my own personal experience, DeepSeek wasn't bad, but it didn't stack up against ChatGPT at the time. And now that gap has widened with the huge progress made by OpenAI’s rivals, Claude and Grok. Both appear to have outpaced even OpenAI's latest ChatGPT v5.0 model. Make no mistake, the US still dominates the globe when it comes to AI, which I think today’s first chart shows.

CHART 1 • Grok surges while DeepSeek implodes

Elon Musk's Grok has exploded with 27% growth in July, leaving rivals in the dust as xAI's aggressive push for AI dominance pays spectacular dividends. The chatbot that promised to be ChatGPT's "anti-woke" challenger now powers X Premium subscriptions at $40 monthly, while Claude and Perplexity manage anaemic single-digit gains and ChatGPT limps along at just 6%.

Most stunning is DeepSeek's 9% collapse – a dramatic reversal for the Chinese startup that shocked Silicon Valley by claiming to match OpenAI's capabilities for just $6 million. After initially sending shockwaves that wiped $589 billion from Nvidia's market cap, DeepSeek's lustre has faded as questions mount about its true costs and capabilities.

Meanwhile, Grok 4's July launch claims to have brought "PhD-level" performance across all subjects, according to Musk, who secured a $200 million Pentagon contract and integration into Tesla vehicles. I personally don’t think it has performed quite so well, but my experience with it has been impressive.

The momentum is also undeniable – xAI just absorbed X in an all-stock deal, while claiming its models outperform GPT-4o and Gemini on key benchmarks. I think these claims are pretty accurate.

Source: Digital Information World

Want the other four? Become a paid subscriber.

---