It’s been another momentous week in AI, with hardware and software developments that continue to reshape the industry landscape.
Meta signalled a major push into humanoid robotics with its Llama-powered initiative, while Perplexity challenged industry leaders with its Deep Research platform, outperforming Gemini Thinking and OpenAI’s latest models on key benchmarks. Figure AI made waves with its Helix robot system that can understand speech and manipulate objects without specific training, driving its valuation to $40 billion. Meanwhile, AWS CEO Matt Garman outlined ambitious plans for enterprise AI adoption, as NVIDIA grappled with market turbulence despite research showing its GPU computing power doubles every 10 months.
Meta Sets Sights on Humanoid Robots and AI
Meta CTO Andrew Bosworth revealed ambitious plans for a new Robotics product group focused on consumer humanoid robots powered by Llama. In a leaked internal memo, Bosworth emphasized how Meta’s existing investments in Reality Labs technologies like hand tracking and material simulation will complement robotics development. Bosworth emphasised the current success of Reality Lavs but stressed that there was still some way to go to fulfil Meta’s ambitions:
On paper 2024 was our most successful year to date but we aren’t sitting around celebrating because know it isn’t enough. – We haven’t actually made a dent in the world yet.
Perplexity Challenges AI Leaders with Deep Research
Perplexity has launched Deep Research, a new service that generates comprehensive research reports by combining advanced web search, reasoning, and coding capabilities. The company claims its solution outperforms competitors, achieving 93.9% accuracy on SimpleQA and surpassing Gemini Thinking, OpenAI’s o3-mini, and DeepSeek-R1 on industry benchmarks. The service, which can complete most tasks in under 3 minutes, is available free to users with tiered access based on subscription status.
OpenAI’s Deep Research entry makes this a particularly competitive space, with their solution offering an autonomous agent that can analyze hundreds of sources and produce comprehensive reports independently. The development signals a shift from AI as a tool to AI as an autonomous assistant.
Figure AI’s Helix Makes Robotics Breakthrough
Figure AI unveiled Helix, a groundbreaking Vision-Language-Action robot system that can understand speech, reason through problems, and handle household objects without specific training. Two Helix robots can even work together to unpack groceries and organize items, trained on just 500 hours of human demonstrations. The company’s valuation has surged to $40 billion, with backing from Microsoft, OpenAI, NVIDIA, and Jeff Bezos. CEO Brett Adcock said:
To bring robots into homes, we need a step change in capabilities. Helix can generalise to virtually any household item… Like a human, Helix understands speech, reasons through problems, and can grasp any object – all without needing training or code.
AWS Charts Enterprise AI Strategy
AWS CEO Matt Garman outlined the company’s 2025 AI priorities, emphasising security and operational excellence while helping customers transition AI workloads into production:
For 2025, our goal is for customers to move AI workloads really into production that deliver great ROI for their businesses. And that crosses making sure all their data is in the right place, and make sure they have the right compute platforms. We think Trainium is going to be an important part of that.
The company is leveraging partnerships with Anthropic and NVIDIA, with Garman revealing that Anthropic will train their next-generation model on AWS’s Trainium 2 chips. He expressed optimism about the Trump administration’s potential to streamline data center development and help maintain US AI leadership.
NVIDIA’s Growth Challenges
New research from Epoch AI shows NVIDIA’s GPU computing power doubles every 10 months, but the chip giant faces mounting competition. DeepSeek’s breakthrough caused NVIDIA’s largest single-day market value loss in history, with shares plunging 17% and erasing nearly $600 billion in value. The company is expanding its Blackwell chip portfolio and Project DIGITS while confronting challenges from OpenAI’s partnership with TSMC to develop custom AI chips.
