Google’s push to re-establish lead in AI boosts investor confidence

Google’s push to re-establish itself at the vanguard of technological research and artificial intelligence this month has driven its stock to a record high while quietening criticism it had fallen behind rivals.

Over the course of December, the Big Tech group impressed investors with a more advanced version of its AI models and applications called Gemini 2.0 which has beaten rivals in benchmark testing.

In a flurry of co-ordinated releases, the company also unveiled a new generation of its custom AI accelerator chip — a Tensor Processing Unit (TPU) called Trillium — which aims to challenge Nvidia’s near monopoly on the market.

Google also added the capability to act on users’ behalf and compile complex research reports — Project Mariner — and answer real-time queries across text, video and audio — Project Astra — including via smart-glasses. And it launched video and image generation models called Veo 2 and Imagen 3.

“The last month has transformed the state of AI, with the pace picking up dramatically in just the last week,” said Ethan Mollick, a professor at Wharton business school and author of a book on the technology, describing Google’s releases, in particular Veo 2, as “astonishing”.

“This isn’t steady progress — we’re watching AI take uneven leaps past our ability to easily gauge its implications,” added Mollick.

Additionally, Google confirmed last week that it had made a breakthrough in quantum computing with a chip called Willow. It can hold “qubits” stable for longer, reducing errors and allowing them to perform useful computations.

The company claims it can complete tasks in 5 minutes that would take conventional supercomputers 10 septillion years, but the elusive technology remains years from commercial application.

In more recognition of its edge on research, in October, Sir Demis Hassabis, founder of Google’s AI research lab DeepMind and his colleague John Jumper shared the chemistry Nobel Prize for predicting the structure of every known protein using AI software known as AlphaFold.

The showcase of technological advances — along with three consecutive quarters of double-digit profit growth — have helped revive parent company Alphabet’s share price.

The stock is up 38 per cent this year and briefly touched a record high of $199.91 this week, giving it a market capitalisation of $2.3tn. Still, a $1tn gap remains to close with Microsoft.

Since the release of ChatGPT in late 2022, Google appeared to squander its early advantage in AI, after incubating the underlying research, in particular when its arch-rival Microsoft partnered with OpenAI. It took Google a year to release its own comparable version, Gemini.

“Alphabet has been under the microscope since ChatGPT was released,” said Tiffany Hsia, a US equity portfolio manager at AllianceBernstein, a shareholder in the company. “Gemini 2.0 and the quantum chip gives investors renewed confidence that they are one of the leading tech powerhouses.”

In a sign of growing confidence, chief executive Sundar Pichai — who had faced some of the harshest criticism in his nine-year tenure for the slow AI rollouts in the spring — challenged his counterpart at Microsoft, Satya Nadella.

“I would love to do a side-by-side comparison of Microsoft’s own models and our models any day, any time,” said Pichai during the DealBook Summit earlier this month. He added with a smile that, besides, “they’re using someone else’s models”. 

While the company demonstrates its technological chops, it must work out how to incorporate these innovations into its consumer and commercial applications without stifling the creativity of its engineers.

Pichai has steadily sought to integrate AI into its search engine, while needing to placate investors concerned that such a move will cannibalise advertising earnings.

The search giant still controls 90 per cent of the market, but for the first time in decades faces real competition from AI-powered products by groups including OpenAI, Anthropic and Perplexity that can provide comprehensive answers rather than links.

Google’s solution so far has been “AI Overviews”, brief answers to queries at the top of its results page. Executives have said the feature is popular, but early evidence suggests users click on Overview ads at a lower rate — down 8 per cent year-on-year in the third quarter, according to research by advertising platform Skai.

Other threats remain. After losing a big antitrust case against its search arm in August, the Department of Justice is seeking to force the sale of its Chrome browser, cancel an exclusive contract to be Apple’s default search engine and share the trove of user data that underpins Google’s proprietary webpage ranking algorithms, ad targeting systems and AI model training.

The company is awaiting the results of another monopoly trial in the US focused on its ad tech business when Alphabet’s other core revenue source could be broken up.

Another potential danger is Elon Musk. The world’s richest man holds sway over president-elect Donald Trump after spending $250mn to help win last month’s US election — gaining the power to influence AI regulation and antitrust enforcement.

Musk’s xAI start-up has also built the world’s largest supercomputer in Memphis, Tennessee, in record time. Nicknamed Colossus, it has networked 100,000 cutting-edge Nvidia graphics processing units — and has plans to expand the data centre 10-fold to 1mn chips — which should help xAI’s chatbot Grok catch up with the competition in 2025.

“Sundar does now seem more confident. Because the ethos of Google is to be perfectionist, we may see product launches at a more meticulous, calculated pace, but we mustn’t be so impatient about it,” said AllianceBernstein’s Hsia. “This is a race, and recent developments show they’re still in it.”