Welcome Back!

I’ve been trying to make sense of Trump’s deal-making with Saudi and the UAE in the semiconductor industry of the past week. There are serious geopolitical, regional and business implications here.

Trump’s maniacal deal-making ambitious hide a lot of fairly serious risks in these moves. Neither does Trump’s trip to the Gulf suddenly transform the Middle East into a global AI powerhouse, like it is often claimed in the PR statements.

Giving AI chip capabilities to countries where human rights are not a priority is also a statement in of itself about the AI leadership of the United States in 2025.

That’s not to say that they aren’t building datacenters. The United Arab Emirates (UAE) and the United States will establish a regional artificial intelligence technology cluster in Abu Dhabi, together with a 5 Gigawatt (GW) capacity data centre, building out short-term 1 GW capacity during phase one.

Trump has now with these deals opened the door for the Gulf to obtain the world's most advanced AI chips. Good in theory for the likes of Nvidia, AMD and others.

The Biggest Deals

This week's U.S.-Saudi Investment Forum in Riyadh showcased a growing partnership in artificial intelligence between the two nations, alongside Saudi Arabia’s launch of Humain, a new AI venture backed by its $940 billion Public Investment Fund (PIF).

• Strategic investment: DataVolt committed $20 billion for U.S. AI data centres, while Saudi’s new AI firm Humain plans to deploy 500 MW each of AMD and Nvidia systems (roughly $20+ billion) over five years, the largest bilateral deal in history.

• Ecosystem expansion: Humain signed a $5 billion accord with AWS to build an “AI Zone” running GenAI services, and a non-binding MOU with Qualcomm for edge-to-cloud AI silicon.

• Broader impact: Trump’s AI advisor David Sacks emphasised building “the biggest partner ecosystem” for AI, specifically naming Saudi Arabia as a crucial strategic partner to “win the AI race and shift the balance of power” in global AI infrastructure.

Humain’s flurry of deals sounds impressive, at least on paper. Huge deals with Nvidia, AWS and so forth.

Humain’s partners:

The deal also involved Saudi buying U.S. weapons.

“Humain will be owned by Saudi Arabia’s Public Investment Fund, and will work on developing AI models as well as building data center infrastructure, according to a press release. Humain’s plans eventually include deploying “several hundred thousand” Nvidia graphics processing units.”

What is Humain again?

Humain, is a subsidiary of Saudi Arabia's Public Investment Fund, aims to establish the Saudi Arabia as an international AI powerhouse by combining state-of-the-art infrastructure, advanced AI models, digital platforms, and human capital development.

• The company plans to build AI factories with a projected capacity of up to 500MW, powered by several hundred thousand of Nvidia's most advanced GPUs over the next five years.

• The first phase of deployment includes an 18,000 Nvidia GB300 Grace Blackwell AI supercomputer with NVIDIA InfiniBand networking.

• Humain claims it will deliver AI-native platforms across four core layers: Infrastructure, Cloud, Data & Models, and Applications.

While the stock market seemed to like Trump’s announcements, there are many risks to this sort of pomp and extravagance, including China AI chip risks.

• In Saudi Arabia, Trump and tech leaders from AMD, Amazon and other companies announced AI-related partnerships worth billions of dollars with a new Saudi-state-backed AI infrastructure startup called Humain. Nvidia said it will ship 18,000 of its cutting-edge AI chips for a 500-megawatt data center being built by Humain.

• In the UAE, Trump and Emirati President Mohammed bin Zayed said the two countries will partner to build the largest AI data center outside the U.S., in Abu Dhabi.

Subscribe now

Read more

Good Morning,

I’ve been thinking a lot again about the efficacy of U.S. export controls around AI chips to China. Requiring Nvidia’s H20s to have a license is almost like funding China’s SMIC and Huawei duo to accelerate their own chip development.

China’s Huawei Technologies is preparing to test its newest and most powerful artificial intelligence (AI) processor as a replacement for Nvidia’s H20s chip.

Nvidia said in mid April that it will take a quarterly charge of about $5.5 billion tied to exporting H20 graphics processing units to China and other destinations. China of course wants a home grown AI chip to replace it and Huawei has approached some Chinese tech companies about testing the technical feasibility of the new chip, called the Ascend 910D.

According to the recent WSJ report, the OEM is reaching major Chinese tech firms to analyze the technical value of the Ascend 910D, before bringing it to the real world. Huawei reportedly plans to begin mass shipments of its 910C AI chip to Chinese customers as early as May, 2025 according to Reuters. This won’t just impact Nvidia sales in China, but China’s own capabilities of becoming more self-reliant in the AI chip sector.

On April 9, the U.S. government told Nvidia it would require a license to export the chips to China and a handful of other countries and the consequences are likely that China has no choice but to accelerate their own SMIC and Huawei efforts, the most capable duo currently in China’s AI semiconductor chip industry.

Is Huawei’s Ascend 910D a game changer?

There are some reports that insinuate people related to the matter say that Huawei sees the new AI chip as more powerful than Nvidia’s H100 and could start sending its first batch of samples by late May. How could that be?

For China having a H100 competitor (a chip competitive in 2024) would be a huge achievement. The Ascend 910D is itself designed to outperform Nvidia’s H100 chip, which has been banned from sale in China since 2022 due to US government export controls targeting China’s technological advancements, particularly in military applications.

For China having a home-grown chip instead of relying on Nvidia’s products, which it has already stockpiled in previous years and through various backdoor foreign entities (Singapore, Malaysia, etc…) is a matter of national security.

Subscribe now

Read more

Hey Everyone,

Amazon has been more or less successfully developed and deployed its own custom AI chips, including Trainium for training large models and Inferentia for inference, to reduce reliance on Nvidia GPUs and gain cost advantages in the cloud. Trainium 2 certainly has some potential.

So what about Google’s Ironwood? Ironwood is Google’s first chip for the age of interference that they announced on April 9th, 2025.

“Age of Inference”

It’s Google’s seventh generation of its custom TPU architecture. The chip, known as Ironwood, was reportedly designed for the emerging needs of Google's most powerful Gemini models, like simulated reasoning, which Google prefers to call "thinking." The company claims this chip represents a major shift that will unlock more powerful agentic AI capabilities. Google calls this the "age of inference."

At Google Cloud Next they announced some interesting things:

• Scheduled to launch sometime later this year for Google Cloud customers, Ironwood will come in two configurations: a 256-chip cluster and a 9,216-chip cluster.

• Google is claiming its Cloud is the only AI-optimized platform. I get the sales pitch: Google says it’s the most powerful TPU accelerator the company has ever built, and it can scale to a megacluster of 9,216 liquid-cooled chips linked together by its advanced Inter-Chip Interconnect technology. In this way, users can combine the power of many thousands of Ironwood TPUs to tackle the most demanding AI workloads. Fair enough.

Subscribe now

Read more

Good Morning,

Apple is making OpenAI’s Stargate datacenter project look downright selfish and small. Some breaking news for you today:

Apple’s commitment to spend $500 Billion in the next four years sounds a bit like the BigTech capex bonanza on AI Infrastructure, but here it’s more about manufacturing. Apple has announced a monumental plan to invest over $500 billion in the United States over the next four years. This investment is aimed at bolstering domestic manufacturing, technology development, and job creation.

Read Apple's Brief

• The U.S. government is leaning hard on tech companies to make more commitments to building their businesses in the country, Apple said its 4 year efforts will create 20,000 new jobs.

• Big projects will include a new factory in Houston, TX, to produce servers that support Apple’s company’s in-house AI effort, Apple Intelligence; doubling the value of Apple’s U.S. Advanced Manufacturing Fund to $10 billion; a new academy in Michigan to train people to work in next-generation factories; and more R&D.

In addition to the new Texas facility, Apple said it also plans to hire around 20,000 employees across the U.S. over the period. It’s not clear how many of those will work at the new Houston site.

Build It

The U.S. technology giant said it would work with partners to launch a 250,000-square-foot server manufacturing facility in Houston to produce servers for Apple Intelligence, its AI personal assistant for iPhone, iPad and Mac computers.

"We are bullish on the future of American innovation, and we're proud to build on our long-standing U.S. investments with this $500 billion commitment to our country’s future,” Apple CEO Tim Cook said in a statement.

Apple, as a major consumer electronics company, relies heavily on production outside of the U.S., including consumers in China.

The major server factory that it will be building will be focused on building machines that can handle AI compute.

• Most of the new hires will be focused on research and development, or R&D, silicon engineering, software development, and AI and machine learning, Apple said.

This investment, offered in the context of Apple's expansion plans, also includes substantial funding for the company's manufacturing academy to be established in Michigan, aimed at training small and medium-sized businesses in AI and smart manufacturing techniques. This is not unlike Microsoft’s datacenter plans in foreign countries that have educational components for local AI development.

Apple appears set to increase its expenditures on R&D significantly. This as news that everyone from Silicon Valley hyperscalers to China BigTech are spending more on AI capex.

Important to note here that Apple will expand data center capacities in several states including North Carolina, Iowa, Oregon, Arizona, and Nevada, indicating a broader approach to enhance its technological infrastructure across the U.S. It’s very much in line with U.S. MAGA sentiment. Stargate is also hypothetically a $500 Billion project, though at least with Apple we know they are good for it in terms of actual free cash flow. Apple's free cash flow for the trailing twelve months (TTM) ending in December 2024 was $98.299 billion.

The context here is very much Apple playing politics to Trump’s demands. The iPhone maker faces pressure from the Trump administration over where it chooses to manufacture its products. Apple assembles most of its products in China.

Apple was clear to state that it remains one of the largest U.S. taxpayers, having paid more than $75 billion in U.S. taxes over the past five years, including $19 billion in 2024 alone.

But while the $500 Billion figure looks big, there’s a good reason for that. Apple said its $500 billion investment plan will include work with suppliers across the U.S. and production of content for its Apple TV+ media streaming service in 20 states, as well as new hires and research and development spending. If you put it that way, that sounds far more feasible.

While Apple Intelligence is not a great product in 2025, what it signals is that Apple is doubling down on how it believes AI will be used within its products and services. It’s worth noting that 20,000 new hires isn’t a lot for a company the size of Apple, especially over a four year time period.

Good Morning,

The Trump Administration might force Broadcom into a quasi-acquisition of Intel, and this is how it might happen. These rumors are in fact part of the pressure from the Trump Administration, where according to the WSJ, Broadcom has been closely examining Intel’s chip design and marketing business, adding that the company had discussed a potential bid with its advisers but would likely only proceed if it found a partner for Intel’s manufacturing business.

Broadcom has interest in Intel’s chip-design business, while TSMC is looking at the company’s factories.

Trump will likely use Taiwan’s precious position to try to get TSMC to do the grunt work of that job. With rumors swirling of a potential deal between Intel and TSMC to allow the former's fabs to manufacture the most advanced chips in America, a report from the Taiwanese press claims that TSMC might acquire a 20% stake in Intel's Intel Foundry business.

If this happens, it basically levels up Broadcom into Nvidia like territory, where it would become a BigTech hyperscaler of the future. Already its market cap is up 90% in the past year where it’s helping the likes of OpenAI to make their own chips.

Intel woes, could be Broadcom’s gains and for TSMC it might come at a cost. Broadcom is particularly interested in Intel's chip design operations, aiming to enhance its own capabilities in the semiconductor market. Concurrently, TSMC is evaluating the possibility of acquiring Intel's fabrication plants, which would expand its manufacturing capabilities significantly.

So this amounts to the U.S. trying to “steal” some of TSMC’s capabilities for itself while weakening Taiwan’s own Silicon Shield. That is, TSMC, the world’s biggest contract chipmaker, has separately studied controlling some or all of Intel’s chip plants, potentially as part of an investor consortium or other structure. TSMC is already spread thin with projects in Japan, Germany and Arizona.

I’m convinced this is the End of Intel as we know it

Subscribe now

Read more

Good Morning,

DeepSeek is uncovering some valuable journalism about other side stories of China’s momentum in its own chip industry, contrary to many of the U.S. influencers on the topic I watch sometimes on YouTube.

Founded in 2016, ChangXin Memory Technologies, also known as CXMT, is a prominent Chinese semiconductor integrated device manufacturer headquartered in Hefei, Anhui.

The company produces dynamic random access memory (DRAM) chips and in the 2030s, I expect it to be a legit competitor of the power-house that is South Korea’s SK Hynix. In fact, a lot of Taiwanese, Japanese and South Korean innovation won’t be able to compete with what China’s semiconductor industry becomes. You are seeing signs of this already in 2025.

Subscribe now

China in fact, has a number of names of watch in legacy chip foundries. Nexchip is among Chinese foundries quickly winning market share in the crucial $56.3 billion industry of so-called legacy or mature node chips made on 28 nanometre technology and larger.

The Memory Chip Maker of the Future: CXMT

Read more

Good Morning,

I noticed a Huawei connection with DeepSeek recently that I think deserves some brief info and discussion.

Last week we found out that Chinese tech giant Huawei appears to have exceeded expectations in 2024, reporting annual revenue of over 860 billion yuan ($118.27 billion). With all the Western sanctions on them, in theory that should not be possible.

Last week there were also some rumblings that DeepSeek R1 Is reportedly running inference on Huawei’s Ascend 910C AI Chips. DeepSeek found they were 60% as good as Nvidia’s famous H100s.

Huawei’s Earnings suggest that their 2024 revenue was a 22% increase in revenue from 2023, and the fastest growth since a 32% increase in 2016, according to CNBC calculations of publicly released figures. While the U.S. sees export controls critical in slowing down China’s capability to keep up in bleeding edge models, Huawei’s role in all of this remains huge.

Huawei still remains a darkhorse of Global Tech

AI Proem

Deep Dive: Is Huawei like China’s Nvidia plus Google?

For this primer piece, I spoke to people at Huawei and ex-Huawei cloud employees and referenced company information and public reports. Please get in touch if you’re interested in the content or spot anything you think can be debatable. I’m always open to connecting and learning…

Read more

a year ago · 12 likes · 3 comments · Grace Shao

For DeepSeek, born in 2023, that its portfolio of AI compute hardware includes Huawei's Ascend AI Chips is of course not surprising.

Subscribe now

Read more

You are reading Semiconductor Reports (a sister pub of Semiconductor Things), where I’ll be sharing some insights on Datacenter, Semiconductor, Nvidia and TSMC news in particular.

In 2025 we are going to see incredible Semiconductor coverage begin to take place on Substack. When I can tune into of Tech Soda or of Semiconductor Business Intelligence, you don’t have to look far.

• TSMC’s revenue in the December quarter rose 38.8% from a year earlier to NT$868.46 billion, while net income rose 57.0% to a record NT$374.68 billion.

The most important is the company's fantastic profits in Q4-24: Operating profit is up 62% YoY and 18% QoQ, driven by the high demand for TSMC AI customers.

High Performance Computing grew 19% QoQ to 53% of total revenue, and surprisingly, Mobile grew 17%. 3nm Revenue grew with an astonishing 44% QoQ, now representing 26% of the total revenue. - , Note.

Judy’s TechSoda provided Q&A information that is also invaluable.

Subscribe to TechSode

TechSoda

Stimulating, Original, Dynamic, and Accessible Tech News and Insights covering Semiconductors, Optical, Digital, and AI Industries!

By Judy Lin 林昭儀

Subscribe to Semiconductor Business Intelligence

Semiconductor Business Intelligence

If your business or portfolio depends on Semiconductors, you want to know what's happening. We are analysing the semiconductor business, the supply chain, and the ecosystem for changes in supply and demand. "Your future might be somebody's past!"

By Claus Aasholm

TSMC is in 2025 becoming one of the most important Tech companies to watch globally. It’s also fashionable now to write historical analyses of TSMC or books about it.

Subscribe to Culpium

Culpium, by Tim Culpan

Culpium: noun [U] — a cure for lack of knowledge, understanding or clarity. *taken by those seeking to improve comprehension of technology, semiconductors, and supply chains especially in Asia. *See your therapist if stupidity persists

With the relative decline of Intel and Samsung in 2024, TSMC has an elevated position in the world.

Read Q4 TSMC EPS Report

• On an earnings call, the company predicted Q1 2025 revenue to increase between 32.5 per cent and 36.7 per cent to $25 billion to $25.8 billion, with its gross margin expected to remain at 57 per cent to 59 per cent.

• Nvidia's most sophisticated AI chip, named Blackwell, is composed of several chips merged using a complex chip on wafer on substrate (CoWoS) advanced packaging technology provided by TSMC.

In US dollars, fourth quarter revenue was $26.88 billion, which increased 37.0% year-over-year and increased 14.4% from the previous quarter.

TSMC’s Historic Q4

TSMC profit rose 57% from a year earlier to a record high, while revenue jumped 38.8%. The firm had forecast fourth-quarter revenue between $26.1 billion and $26.9 billion.

Despite reports from Taiwan media that Nvidia was reducing CoWoS-S orders from TSMC, potentially impacting the Taiwanese chip foundry's revenue, Nvidia's Blackwell chips continue to sell as quickly as TSMC can manufacture them.

Another Newsletter I find super informative is the following:

Subscribe to SemiVision

SEMI VISION

We are engineers from Taiwan's semiconductor industry with profound expertise in advanced processes and advanced packaging.

By SEMI VISION_TW

• How is the stock doing? TSMC's Taipei-listed shares rose 12% in the fourth quarter, taking its 2024 gains to 81% amid robust demand for the foundry's advanced chips. On the TPE it’s up 90% in the past year.

• High Performance Computing: HPC revenue was up 19% from the previous quarter.

“The surging demand for AI chips has exceeded expectations in Q4,” Brady Wang, associate director at Counterpoint Research

• In the fourth quarter, shipments of 3-nanometer accounted for 26% of total wafer revenue; 5- nanometer accounted for 34%; 7-nanometer accounted for 14%.

If reading in Traditional mandarin is your thing:

Subscribe to Grab a Chip

Grab a Chip｜半導體速報

晶片知識，輕鬆品味！

By Andy Hsieh

• The Taiwan-based company first released its December revenue last week, bringing its annual total to NT$ 2.9 trillion — a record-breaking year in sales since the company went public in 1994.

TSMC has never been so important on the global stage with the datacenter acceleration and Capex of BigTech and China exploding in the mid 2020s.

Counterpoint’s Wang forecasts 2025 to be another strong year for TSMC, with significant revenue growth fueled by strong and expanding demand for AI applications, both in diversity and volume, according to CNBC.

TSMC’s number are hard to overstate in scale: Gross margin for the quarter was 59.0%, operating margin was 49.0%, and net profit margin was 43.1%.

TSMC’s History is Being more widely read

TSMC’s history is also becoming more popular among Western and english readers.

Construction Physics

Morris Chang and the Origins of TSMC

If you were making a list of the most important companies in the world, you’d find Taiwan Semiconductor Manufacturing Company, or TSMC, somewhere very close to the top. It’s one of a tiny handful of companies, along with Intel and Samsung, that can make the most advanced microchips, and it has a…

Read more

a year ago · 247 likes · 20 comments · Brian Potter

Judy’s Q&A coverage is truly invaluable for understanding TSMC’s balanced point of view.

• TSMC Chief Executive C.C Wei said he believes that U.S. export controls on AI chips for China were a manageable issue for the company.

• In recent years, TSMC has expanded its operations beyond Taiwan to enhance its global footprint and meet the increasing demand for semiconductor manufacturing. In Japan, Arizona, Germany and so forth.

Read TSMC Fourth Quarter Details

Guidance 2025:

Based on the Company’s current business outlook, management expects the overall performance for first quarter 2025 to be as follows:

• Revenue is expected to be between US$25.0 billion and US$25.8 billion; And, based on the exchange rate assumption of 1 US dollar to 32.8 NT dollars,

• Gross profit margin is expected to be between 57% and 59%;

• Operating profit margin is expected to be between 46.5% and 48.5%.

TSMC pioneered the pure-play foundry business model when it was founded in 1987, and has been the world’s leading dedicated semiconductor foundry ever since.

It’s hard to put into words how much some of the world’s most valuable companies like Apple or Nvidia depend upon TSMC and their partners. This makes Taiwan the most important technology nexus in the world by a wide margin in my humble opinion.

TSMC’s efficiency and world leading position along with its ability to adapt and innovate in response to market needs is unparalleled in the semiconductor industry. This means more Western companies are going to build satellite teams in Taiwan itself in the years ahead.

Capex is also increasingly quickly mirroring the unique opportunity of the times. TSMC, which is building new fabs in the United States, Japan, Germany and Taiwan, said it expects its capital spending for this year to be between $38 billion and $42 billion, an increase of as much as 41%.

Hey Everyone,

It’s hard to put into context the growth and earnings performance we are seeing out of Nvidia. Nvidia’s stock is also lifting semi stocks (entire ETFs, the NASDAQ, etc…) and investor confidence in the beneficiaries around what’s driving Nvidia’s revenue growth.

Nvidia reported results that beat on both the top and bottom lines, with sales climbing 94% year-on-year to $35.08 billion. How can we fully appreciate those numbers at that kind of scale? It makes the rest of the magnificent seven look like value stocks in comparison.

The insane GPU demand means Blackwell is a sure-fire winner. You cannot really understand today’s Semiconductor industry without understanding the rise of Nvidia and this peak demand we are seeing in the mid 2020s.

Nvidia’s datacenter growth showed revenue of $30.8 billion, up 17% sequential and up 112% year on year. Nvidia is the GPU company of the world in 2024, there really isn’t any worthwhile competition.

Nvidia has largely cornered the market for the high-powered chips driving the world’s most advanced artificial intelligence models and even the likes of xAI and Meta are betting big on bigger and better datacenters. If diminishing returns isn’t coming from the actual models, or the world running out of quality data, the datacenter competition means fewer companies can make state-of-the-art frontier models in 2025.

The AI chip is changing the entire trajectory of technology and BigTech stocks. Nvidia and its supplier TSMC, are at the center of the Generative AI era. While some make some interesting contrarian arguments against Nvidia’s dominance, it’s not entirely convincing considering Nvidia’s place in the AI chip market today. Nor is the argument that companies like Amazon can make a dent in their market dominance rational or realistic.

Let’s look at some of Nvidia’s Earnings numbers and why it’s hard to even acknowledge completely what has occured in a relatively short period that included the emergence of ChatGPT, Generative AI and the titanic capex spend by BigTech and Cloud players in AI chips.

Read more

Hey Everyone,

It’s taking me a long time to get this Newsletter of the ground and running. Basically I want to create deep dives about the semiconductor industry, but I have a lot to learn. I also have some ideas on how this all plays out.

AMD announces New Chip

AMD’s Instinct MI325X’s rollout will pit it against Nvidia’s upcoming Blackwell chips, which will start shipping in significant quantities early next year.Blackwell is sold out for a long period.

• It’s not clear if AMD’s AI chips can be one day seen by developers and cloud giants as a close substitute for Nvidia’s products.

• The Instinct MI325X, as the chip is called, will start production before the end of 2024.

AMD’s Advancing AI event

At its "Advancing AI" event in San Francisco, AMD announced:

Watch the entire event:

• Its latest server chip, code-named Turin, which AMD CEO Lisa Su said offers significant performance benefits over Intel's latest Xeon processors.

• The MI325X, an updated version of AMD's Instinct GPU, where it competes directly with Nvidia. AMD also previewed the MI350, which it expects to ship in the second half of next year.

• Other semiconductors for faster networking, as well as a version of its PC chip designed to power the first Copilot+ PCs designed specifically for business use.

Subscribe now

The company showed off its new 5th Gen AMD EPYC central processing units (CPU) for servers, provided details on its Instinct MI325X AI accelerator, and showed off its Ryzen AI PRO 300 processors for AI PCs for enterprise customers.

Share

The relative decline of Intel will certainly help the rise of AMD. But can they ever truly compete with Nvidia on AI chips?

I may go into more detail about AMD’s new AI chip in a separate shorter article. In the meantime I’m also considering covering breaking news in the space, to get my feet wet. While Semiconductor Things goes through some of the news in the industry twice a month, a lot of key events I want to write about may end up here.

The Great Datacenter Boom is about to accelerate A.I. Infrastructure globally

Read more

Hey Everyone,

This publication used to be about AI startups and wants to be about the future of the semiconductor industry with deep dives. But sometimes those two topics sort of converge. So this piece is going to be more like a stub, or footnote.

This week Rain AI’s movements have come to light.

Subscribe now

Read more

Hello Everyone,

This is an announcement about the future of this Newsletter. Unfortunately not enough people signed up for the paid version of AI Startups and Funding, so this Newsletter is going to pivot into deep dives around the Semiconductor and AI chip industry with a focus on Taiwan.

I will still blog about AI startups and funding on this Newsletter however on occasion. If you are a paid subscriber please unsubscribe and carefully evaluate if you want to support me going forward and this may not align with your interests.

I’m having to pivot my Newsletter strategy to survive as a writer on the future of technology. Ultimately it’s my readers who decide which direction I go. Buying a domain on whim is risky, but these are geopolitically sensitive times and there’s an odd urgency here in the space.

Why AI Chips and Semiconductor Industry?

With connections to Taiwan, I feel it’s important to tell more of TSMC and related stories as I learn about this space with my readers.

Semiconductor Reports ™ is born out at a time of much semiconductor and datacenter geopolitical upheaval, investment and expansion. It’s an exciting period that I want to more fully cover.

This is also due to the success of my recent entry into the Semiconductor Newsletter space, namely:

Semiconductor Things.

Semiconductor Things™

This Newsletter was built to solve the pain point of getting the latest news on A.I. chips, semiconductors, datacenter innovation and chip news.

By Michael Spencer

So the idea here is one is a lower tier rundown of weekly news and this one is more occasional but deep dive reports that are highly valuable. The difference is a $5 to a $19-20 price range.

• Semiconductor Things (more rundown style and current affairs)

• Semiconductor Reports (higher tier deep dives, more hours of research on my part)

So there you have it, I’m branching out from emerging tech and AI into more the more datacenter and AI chip complex semiconductor space.

For those of you interested in emerging tech, I’m trying to create a platform of Newsletters that’s informative and helps you invest in thinking more about the future.

What kinds of articles?

The cadence and post frequency is yet to be determined, but likely won’t be that often, given the effort involved in each piece.

• Semiconductor Deep Dives

• Historical perspectives

• Datacenter and AI chip startups

• Deep dives on breaking stories

Subscribe now

Hello Everyone

Technically I’m on break until May, but yet another Paris based ex-Google DeepMind startup has gotten considerable funding. We talked about them earlier in the year as part of the rising Paris Generative AI scene for startups, especially those formerly at Google DeepMind.

Insider and others report that Holistic AI has gotten over $200 million in funding. This company operates in a really emerging and important niche. This is not the same Holistic AI that does AI risk and safety products based in the UK, but a startup working primarily on agentic AI foundational models.

Back in January, 2024 it was reported that a pair of scientists at Google DeepMind, Laurent Sifre and Karl Tuyls, are reportedly in discussions with potential investors to establish their own artificial intelligence (AI) startup in Paris. It seems those talks are about to conclude nearing May, 2024 and it means they will become a major AI startup in Paris, bolstering already a good lineup of new companies there.

This also signals yet another team within Google DeepMind that have gone their own way, as Bloomberg has signaled the move around 5 months ago.

They are like the Mistral of AI regulation in a sense. Mistral also got huge funding out of the blue and have made a name for themselves in the Generative AI open-source (open-weight) space.

The team at Holistic have two co-CEOs, Charles Kantor and Karl Tuyls. Besides the scoop by Insider, they have not official come out of stealth or done PR on the raise, so it may not be fully finalized.

Holistic will also be focused on new multi-agent AGI, according to investor sources. That is they want to build and generate frontier multi-agent and multi-modal foundational models. Agentic AI is one of the buzzwords without much to show from the likes of Cognitive Labs, Magic AI d others who are also flirting with the AGI terminology. It’s not yet clear what agentic AI properties OpenAI’s GPT-5 might have.

This $200 million Series A is large enough to allow them to scale to become a leader in this nascent industry. More details on the funding round should emerge soon.

The round should reach closer to $220 million. Both scientists are considered leaders in the field, the report said. Sifre co-authored groundbreaking 2016 DeepMind research on the capabilities of AI systems in the game of Go. Tuyls has researched game theory and multi-agent reinforcement learning. 

Holistic is in a position where they could collaborate with Mistral, Kyuntai and others to build France’s overall Generative AI competitiveness and for the EU. The UK and Germany are not exactly thriving in this new nascent space so far in the early 2020s. Will Holistic become the ‘Mistral of Agentic AI’ is the question?

Hello Everyone,

Given advances in robotics and Generative AI, I believe a new wave of robotics startups and funding for advanced robotics is near.

There is one recent event here that’s worthy of mentioning and I’d like to break it down a bit. We are seeing a wider range of hardware and software approaches to this “Robot Renaissance” that is likely coming in the years ahead.

Here I am referring to Collaborative Robotics. The way robots collaborate in the future is going to be a little bit different. This major new funding occured on April 15th, 2024.

General Catalyst is super involved. The Series B is quite large. The $100 million Series B, led by General Catalyst and featuring Bison Ventures, Industry Ventures and Lux Capital. That brings the Bay Area firm’s total funding up to $140 million. General Catalyst’s Teresa Carlson is also joining the company in an advisory role.

Read blog

“Enabling Pragmatic, Trustworthy Automation”

Subscribe now

In a geopolitical environment where logistics and supply chains are being disrupted, what role will automation and robotics play? GC’s Global Resilience thesis focuses on modernizing critical sectors - defense, industrial, energy, and healthcare. In industrial, we see constant pressure to do more with less. They say, companies need to increase output, reduce lead times, and remain cost-competitive while maintaining quality and moving production closer to home.

Their founder is Brad Porter.

As the two-year-old startup’s name implies, Collaborative Robotics (Cobot for short) is interested in the ways in which humans and robots will collaborate, moving forward.

While a “ChatGPT moment for robotics” is likely years away, there are some promising Robotics startups forming in the early to mid 2020s.

For example, in March 2024 robotics investments total $642M.

Hyperscale talent are’t just pivoting into Generative AI, but into new kinds of software that will lead to a Machine led economy.

Cobot has the pedigree, as well, with staff that includes former Apple, Meta, Google, Microsoft, NASA and Waymo employees. Porter himself spent more than 13 years at Amazon.

Robotics Funding March, 2024

Subscribe now

Read more

Hello Everyone,

I’ve been giving more research and thought into the AI datacenter explosion’s energy and water problem.

It was Earth day and something is a bit off in Silicon Valley. Microsoft is putting PR around Energy dynamics of datacenters that looks like this in Nature.

Sam Altman and Microsoft seem to mirror each other here as we have seen in previous investments in Nuclear Fusion. This is highly problematic for antitrust probes for OpenAI’s future. But the story continues.

April 22nd, 2024. Exowatt, a next-generation renewable energy company, today unveiled a first-of-its-kind modular energy platform designed to power energy-intensive data centers along with a $20 million seed round from a16z, Atomic, and Sam Altman. As the boom in AI technology escalates both the demand and the energy costs for data centers, Exowatt's solution arrives at a critical time.

Sam Altman, happy 39th birthday.

We’re entering an era where the Cloud is going to scale into AI datacenters that might be significantly more power and water hungry that we can imagine today in the 2020s.

Subscribe now

Indeed, AI could consume up to 3.5% of the world’s electricity by 2030, Gartner predicts.

Gartner is predicting that by 2030, AI could consume up to 3.5% of the world’s electricity. For context, global energy usage tallied approximately 25,500 terawatt-hours in 2022, according to Statista. So what startups in Nuclear Fusion or elsewhere are going to solve this issue and meet this demand?

Read more

Hello Everyone,

This is going to be a different kind of post. G42 was founded in 2018, so it’s really one of the most impressive AI startups in the Middle East. Now it’s almost like a vassal company of Microsoft, at least on paper with this new deal. G42 has very strong ties with the UAE power structure.

Today I’m obsessed with this huge move in AI funding that has geopolitical and AI infrastructure implications for the future of AI. Microsoft is setting up conditions that they are among the big winners of hyperscale AI.

G42 had ties with China but were forced to divest from China to allow this deal to go ahead. This deal is almost like a bribe to secure the UAE’s involvement with Sam Altman’s datacenter infrastructure global coalition.

Microsoft will invest $1.5 billion in United Arab Emirates-based artificial intelligence firm G42, giving the U.S. tech giant a minority stake and a seat on the board, which will of course be filled by Brad Smith.

Brad is positioning this as enhancing the UAE’s position, but really it’s divide and conqueror for the U.S. on the global stage in the AI arms race, and I think it’s an existential risk for global conflict in the geopolitics of AI.

Microsoft in their PR attempts to speak about it in agonistic terms:

• This strategic investment will enhance the UAE’s position as a global AI hub and provide further opportunities for partners and customers to innovate and grow.

• G42 and Microsoft will collaborate to ensure the benefits of secure AI technologies and cloud capabilities are responsibly shared with growing economies globally. 

• Microsoft’s Vice Chair and President Brad Smith will take a seat on the G42 Board.

• Both companies will support the establishment of a $1B fund for developers.

• The partnership includes a first of its kind Intergovernmental Assurance Agreement as both parties commit to world-leading standards in AI safety and security.

But the truth is this has significant geopolitical implications that positions Microsoft in the crossfire between the U.S. and China. I would not be surprised if this impacts Microsoft Research Asia in the long-term.

Subscribe now

Read more

Image: Prompt for ChatGPT: "An abstract image showing the concept of strength in diversity with natural elements"

Hey Everyone,

The AI Startup I’ve been thinking the most about this week is likely Hume AI.

Meet Hume’s Empathic Voice Interface (EVI), the first conversational AI with emotional intelligence.

With a single API call, interpret emotional expressions and generate empathic responses. Meet the first AI with emotional intelligence.

Try it here: demo.hume.ai

Hume's EVI API can do more than just chat - it can also take actions!

EVI will be publicly available in April. If you’re a developer interested in earlier access to the API fill out this form: https://share.hsforms.com/15hCR14R4S-e-dlMwN42tkwcjsur

Subscribe now

Series B

They have just raised $50 million in a Series B, and a NY based.

The round was led by EQT Ventures, with participation from Union Square Ventures, Nat Friedman & Daniel Gross, Metaplanet, Northwell Holdings, Comcast Ventures, and LG Technology Ventures.

Those are some slightly unusual backers.

• I am very bearish on facial recognition and emotive prediction in general. Hume AI is a startup and research lab building artificial intelligence optimized for human well-being.

• Hume operates at the intersection of artificial intelligence, human behavior, and health and well-being and has created an advanced API toolkit for measuring human emotional expression that is already used in industries spanning from robotics to customer service, healthcare, health and wellness, user research, and more.

The measuring of emotions by AI is not without controversy but Hume AI brands itself as “the feelings lab”.

Read their Docs

In the PR they say that Hume AI was founded by Dr. Alan Cowen, a former Google researcher and scientist best known for pioneering semantic space theory – a computational approach to understanding emotional experience and expression which has revealed nuances of the voice, face, and gesture that are now understood to be central to human communication globally.

His tagline is typically “teaching AI to make people happy”.

A Voice AI with Emotional Intelligence?

Left to right: Michael Opara and Lauren Kim founding engineers; Alan Cowen, founder, CEO and chief scientist; and Janet Ho, COO, are building "emotionally intelligent" AI models.

Capabilities of EVI in March, 2024

EVI understands the user’s tone of voice, which adds meaning to every word, and uses it to guide its own language and speech. Developers can use this API as a voice interface for any application.

✨EVI has a number of unique empathic capabilities ✨

1. Responds with human-like tones of voice based on your expressions.

2. Reacts to your expressions with language that addresses your needs and maximizes satisfaction.

3. EVI knows when to speak, because it uses your tone of voice for state of the art end-of-turn detection.

4. Stops when interrupted, but can always pick up where it left off.

5. Learns to make you happy by applying your reactions to self-improve over time.

And includes fast, reliable transcription and text-to-speech and can hook into any LLM.

Over the years the team at Hume AI has grown as well:

See Publications

Subscribe now

Voice AI’s Future?

The company believes AI voice products have the ability to revolutionize our interaction with technology; however; the stilted, mechanical nature of their responses is a barrier to truly immersive conversational experiences. The goal with Hume-EVI is to provide the basis for engaging voice-first experiences that emulate the natural speech patterns of human conversation.

That’s sounds somewhat like what ElevenLabs have been doing.

Their demo is voice driven so you better have a microphone connected:

Try Demo

This is the sort of startup that someone like Amazon might want to acquire one day.

Hume is a research lab and technology company. Our mission is to ensure that artificial intelligence is built to serve human goals and emotional well-being.

I just wonder if the prediction of emotion actually serves emotional well-being? The very act seems like an API begging to be abused.

Their empathetic voice interface, EVI, is an emotionally intelligent conversational AI is the first to be trained on data from millions of human interactions to understand when users are finished speaking, predict their preferences, and generate vocal responses optimized for user satisfaction over time. These capabilities will be available to developers with just a few lines of code and can be built into any application.

That sounds like the foundation of a social commerce platform AI. Valuable to E-commerce leaders like Amazon, Pinduoduo, Walmart, Alibaba and others.

The company’s eLLM enables EVI to adjust the words it uses and its tone of voice based on the context and the user’s emotional expressions. Does that suggest it’s multimodal and require video as well?

Hume AI seeks to understand more nuanced and often multidimensional emotions of its human users. On its website, the startup lists 53 different emotions it is capable of detecting from a user, including:

1. Admiration

2. Adoration

3. Aesthetic Appreciation

4. Amusement

5. Anger

6. Annoyance

7. Anxiety

8. Awe

9. Awkwardness

10. Boredom

11. Calmness

12. Concentration

13. Confusion

14. Contemplation

15. Contempt

16. Contentment

17. Craving

18. Desire

19. Determination

20. Disappointment

21. Disapproval

22. Disgust

23. Distress

24. Doubt

25. Ecstasy

26. Embarrassment

27. Empathic Pain

28. Enthusiasm

29. Entrancement

30. Envy

31. Excitement

32. Fear

33. Gratitude

34. Guilt

35. Horror

36. Interest

37. Joy

38. Love

39. Nostalgia

40. Pain

41. Pride

42. Realization

43. Relief

44. Romance

45. Sadness

46. Sarcasm

47. Satisfaction

48. Shame

49. Surprise (negative)

50. Surprise (positive)

51. Sympathy

52. Tiredness

53. Triumph

I’m not clear why the product is named after the philosopher David Hume. David Hume had a sharp wit and was a skeptic, certainly not the empathetic sort.

"What sets Hume AI apart is the scientific rigor and unprecedented data quality underpinning their technologies," said Andy Weissman, managing partner at Union Square Ventures.

They do have a lot of publications but I’m still unclear on exactly what they do, even after reading their PR, marketing material and browsing their papers lightly.

What are the Applications of EVI?

This feels like something that could automate screening job applicants and HR duties. Who needs an empathetic LLM? Perhaps HR, sales and customer service for the most part.

The EVI was built using a kind of multimodal generative AI that combines standard large language model capabilities with expression measuring techniques. The company calls this novel architecture an “empathic large language model” or eLLM, and says this is what allows EVI to adjust the words it uses and the tone of its voice, based on the context and emotional responses of human speakers.

It’s clever, with the ability to accurately detect when the speaker is ending their conversational turn, so it can start responding almost immediately, with latency of less than 700 milliseconds. It’s sensitive too, and it will stop speaking if the user interrupts it. It all adds up to a much more fluid, humanlike conversational interaction.

However, I am not clear on their case studies.

Read the paper

As a Series B startup they are relatively early still in their application of their technology. They have around 35 employees so far but with the new funding they should be able to expand their team and find actual applications.

They envision a future where AI systems use scientific approaches to fulfill human needs. Emotional intelligence is the missing ingredient needed to build AI systems that proactively find ways to improve your quality of life. For me ideas are:

• Senior care

• Early childhood education tutors

• Some nursing interactions

• Some business development scenarios

• Some customer success circumstances

• For robotics and robot to human interactions

• For smart car intelligence, “you sound sad today, want me to play you your favorite song?”

Where is emotional intelligence needed in society and in day-to-day situations in modern life? There are a lot of potential interactions here.

Their series A was only a bit under $13 million.

"Hume AI’s toolkit supports an exceptionally wide range of applications, from customer service to improving the accuracy of medical diagnoses and patient care, as Hume AI’s collaborations with Softbank, Lawyer.com, and researchers at Harvard and Mt. Sinai have demonstrated."

The influx of new funding values the startup at $219 million. Nat Friedman was involved, but wait, he’s involved in nearly everything!

The company also announced the launch of “Hume EVI,” a conversational voice API that developers integrate into existing products or build upon to create apps that can detect expressional nuances in audio and text and produce “emotionally attuned” outputs by adjusting the words and tone of the AI.

Hume AI TL;DR

Hume AI's EVI: A leap forward in AI chatbot technology

1. Hume AI's EVI is a unique voice interface currently undergoing beta testing.

2. Unlike other AI chatbots, EVI can interpret the emotional tone of human speakers to better comprehend their speech.

3. It adjusts its responses based on the user's emotional tone, using data from numerous human interactions to create a proper vocal response almost instantly after the user has finished speaking.

I’ve been writing about Voice tech and Voice AI for a long time. It’s been a long wait to HER (2013). Will corporate AGI feel like a smart Operating system? Possibly, possibly not so emotionally intelligent.

Besides its empathic conversational capabilities, EVI supports fast and reliable transcription and text-to-speech functionality, meaning it can adapt to a wide range of scenarios. Developers will be able to enhance it even further by integrating it with other LLMs.

If I bought a robot for $12,000 in 2026, you’d think it would need some emotional intelligence capabilities.

The founder and CEO is just 33, I think his team have time to perfect this product. These capabilities will be available to developers with just a few lines of code and can be built into any application. Maybe this is the Stripe of Empathetic LLMs.

If it feels a bit speculative and nebulous, we have to remember the startup was founded just recently in 2021.

EVI uses a new form of multimodal generative AI that integrates large language models (LLMs) with expression measures, which Hume refers to as an empathic large language model (eLLM). Time will tell if this API based product can deliver in real-world settings.

Read the blog

Originally known as the Startup that teaches AI to read emotions. Thus far I think they’ve seen the most traction in pilots involving healthcare and patient facing chatbots. Hume’s AI has been integrated into applications in industries like health and wellness, customer service and robotics, Cowen said.

• Healthcare

• Robotics

• Beauty & Wellness

• Customer Service

Hello Everyone,

One of my favorite AI startups specifically in Generative AI in healthcare has to be Hippocratic AI.

I wrote about them back in May, 2023:

AI Supremacy

Hippocratic A.I. is First Safety Focused LLM for Healthcare

Source: Pexels Welcome Back! Today I want to introduce to you an important new startup in the Generative LLM space. 📊 Milestone Reached 20,000 My goal for 2022 on Substack was to reach 20,000 readers, and this publication with your serenity is outperforming that target. Today May 17th (which happens to be my birthday) I will reach that milestone. This me…

Read more

3 years ago · 17 likes · Michael Spencer

They recently announced a partnership with NVIDIA to co-develop super low-latency “Empathy Inference.” Empathy inference will help patients to naturally build an even deeper emotional connection with Hippocratic’s voice-enabled AI agents—a vital component of the patient-provider experience. 

Subscribe now

Do No Harm

Munjal Shah, Hippocratic AI’s co-founder and CEO, emphasizes its safety-centric approach. On March 18th, at Nvidia’s GTC, NVIDIA Healthcare launched Generative AI microservices to advance drug discovery, MedTech and digital health and I was happy to see Hippocratic AI was involved.

Their Series A raised $53 million in funding with a valuation of $500 million.

The round was co-led by Premji Invest and General Catalyst with participation from SV Angel and Memorial Hermann Health System as well as existing investors Andreessen Horowitz (a16z) Bio + Health, Cincinnati Children’s, WellSpan Health, and Universal Health Services (UHS).

Hippocratic has raised $120 million to date, including the $50 milion last may.

Nvidia Connections here are Strong

Munjal Shah, a serial entrepreneur, founded Hippocratic AI along with a group of physicians, hospital administrators, healthcare professionals and artificial intelligence researchers from El Camino Health, Johns Hopkins, Washington University in St. Louis, Stanford, Google, and Nvidia. 

Sandesh Patnam, Managing Partner at Premji Invest, praised Hippocratic AI’s safety-first stance, aligning with the Azim Premji Foundation’s ethos of societal value creation. He expressed optimism about their potential to revolutionize healthcare with a secure LLM.

Autonomous Telehealth Nurses?

Importantly Nvidia announced a collaboration with Hippocratic AI recently, in digital nursing. Hippocratic mentions how much cheaper this might be. In the area of one tenth of the cost, where its cheap AI agents that offer medical advice to patients over video calls in real-time. This received some criticism from the press. However I find it very intriguing.

Reducing healthcare costs is crucial as populations age with a lack of nurses and poor doctor to patient ratios in many places. $9 instead of $90 an hour in some cases.

“Voice-based digital agents powered by generative AI can usher in an age of abundance in healthcare, but only if the technology responds to patients as a human would,” said Kimberly Powell, vice president of Healthcare at NVIDIA in a press release recently.

This means Hippocratic AI is swinging big in some areas and has some potential in the Telehealth industry.

Hippocratic AI is hugely ambitious here if it’s seeking to automate some of the tasks that Nurses might do. However the TAM of automation in healthcare is also high, even to alleviate some of the burden on clinicians like nurses. Nurses are dealing with worker shortages, that predate the covid-19 pandemic, which Hippocratic seeks to address.

Nurses may not actually make us much as Hippocratic AI has stated in its marketing material, but I think we get the point. Clinicians and especially physicians and nurses do need help and some AI tools might help in an era of more digital friendly Telehealth interventions.

Hippocratic AI’s LLM is trained on evidence-based content and utilizes a unique “constellation architecture” with multiple models working together to improve accuracy and minimize errors. Additionally, human supervision is built-in when necessary.

• Per the company, this platform enables health systems, payors, and other stakeholders to "hire" AI agents capable of completing low-risk, non-diagnostic, patient-facing tasks.

• Its product of AI agents is actually fairly pragmatic: This marketplace is part of the company’s vision around how generative AI can provide super staffing and healthcare abundance and equity for the first time in history.

• Initial roles include chronic care management, post-discharge follow-up for specific conditions such as congestive heart failure and kidney disease, as well as wellness and social determinants of health surveys, health risk assessments, and pre-operative outreach. However, these agents won’t be allowed to speak with patients unsupervised until phase three of the safety testing is completed.

This is also a profitable pragmatic approach to improving patient-care with AI. I don’t find the media criticism fully justified when seniors in more developing countries are going to be rapidly aging with fewer bodies to take care of them than ever before in human history.

Early Use Cases

• AI agents for chronic care management

• Post-discharge follow-up

• Congestive heart failure management

• Kidney disease management

• Pre-operative outreach

Can this technology one day take some of the tasks of nursing assistants or medical office assistants? For sure. But that might allow them to focus on more person facing tasks. The point is to reduce healthcare costs that risk spiraling out of control in the next decades.

Hippocratic AI to me seems like solutions built for the real world.

Co-founded by CEO Munjal Shah, alongside a group of physicians, hospital administrators, healthcare professionals, and artificial intelligence researchers from El Camino Health, Johns Hopkins, Washington University in St. Louis, Stanford, Google, and Nvidia, Hippocratic AI is advancing a safe focused Large Language Model (LLM) for healthcare.

Generative AI-powered staffing marketplace

When you think of a company like Teladoc, you wonder why they never invented something like this. Hippocratic AI is truly a pioneer now.

That Nvidia is backing them also just adds credibility.

Hippocratic AI says:

These AI healthcare agents will augment the human staff of various health organizations—from health systems to payors to digital health companies to pharma—to complete complex, patient-facing, yet non-diagnostic tasks over the phone. They will be able to deliver seamless, personalized, and conversational patient interactions, working to mitigate staffing shortages while enhancing access, equity, and patient outcomes.

For a lot of seniors, Telehealth interventions make more sense than phone calls. Especially as the Digital AI agents become more sophisticated and empathetic.

The startup also has a handful of health system investors-- Memorial Hermann Health System, Cincinnati Children’s, WellSpan Health, Universal Health Services (UHS), HonorHealth and OhioHealth. So Hippocratic AI is rooted in the actual industry it’s trying to serve.

Testing, Transparency and Pilots

The company has already completed the first two phases of safety testing, involving over 1000 nurses and 130 physicians who interacted with the AI agents and assessed them on various safety measures. The results, published in a paper titled “Polaris: A Safety-focused LLM Constellation Architecture for Healthcare”, show promising outcomes. For instance, nurses rated the AI’s ability to educate patients about their condition at 89.82%, compared to a human nurse’s average score of 80.64%.

The scope for Hippocratic AI’s products is actually fairly broad.

The series A round was significantly oversubscribed, allowing the company to select lead investors who were patient enough to allow the company to prioritize safety over short-term revenue and profit, according to executives. That they have managed to get $120 million so fast is a credit to their diligence and network.

Hippocratic said it will use the new capital to fund these safety tests and to accelerate further product development, executives said. The discrepancy between their rigorous emphasis on safety and the criticism in the media for “automating” or “undercutting” nurses is truly fascinating to me. Nursing burnout is a real problem and AI agents can help more than they can hinder.

Polaris: A Safety-focused LLM Constellation Architecture for Healthcare

It looks like their paper came out just hours ago (5 hours to be exact):

Read the Paper

Polaris

The Hippocratic AI agent was even part of Jensen Huang’s Keynote, can you see the snippet of video here.

In terms of automating routine patient interactions I think their framework has a lot of potential.

According to the company, subjective measures, assessed against human nurses, revealed high levels of comfort and effectiveness in interacting with Hippocratic AI's agents.

On Autopilot?

The staffing marketplace enables health systems, payers, and others to “hire” auto-pilot generative AI-powered agents to conduct low-risk, non-diagnostic, patient-facing services to help address the massive shortage of healthcare nurses, social workers and nutritionists in the U.S. and worldwide. 

Do people feel more comfortable interacting with AIs? Sometimes the answer is yes.

• Nursing

• Social Work

• Nutritionist

Does it lessen the repetitive task burden on these professionals? Probably yes.

Generative AI and agents certainly will become a bigger part of our healthcare systems in the decades ahead and Hippocratic AI may contribute something to the field it appears. They believe they have the evidence to prove it.

Their variety of specialist support agents is very impressive! See page 17 of the paper.

Study participants rated the AI agent on par with U.S. licensed nurses on multiple dimensions. And their systems will only get better over time and with more experience. It’s not clear if this could lead to any job losses in medical administration or nursing assistants or other roles, but as Generative AI gets more involved in the healthcare industry of course it could eventually.

It’s incredible to see how far they have come in such a short time, ten months ago here is an interview the CEO did with Bloomberg:

Suffice to say that Hippocratic AI have a lot more to be confident about heading further into 2024.

The healthcare industry as a whole has been more reluctant to adopt the emerging AI trend due to patient safety concerns. Eventually is has to welcome emerging tech to off-set rising Healthcare costs and clinician staffing concerns and automate some of the admin involved. This is really important as Boomers and GenX are retiring and the population pyramids of the world change radically.

Hey Everyone,

It’s been an incredible week for robotics innovation. And it just got even better! Can you imagine a seed round of $70M to Build AI-Powered Robots for Any Application?

This is one of the most massive seed rounds for robotics-AI we’ve ever seen.

A huge Google and UC Berkeley team spin-off is now taking place with a new company called Physical Intelligence, which is breaking news. Before we talk about Physical Intelligence and their funding round though, one quick thing:

OpenAI is really investing in the future of robots here, basically out-doing Tesla itself. It’s contributing to a better robot-human interface. OpenAI has partnered with Figure AI, a startup developing humanoid robots.

With OpenAI, Figure 01 can now have full conversations with people

• OpenAI models provide high-level visual and language intelligence

• Figure neural networks deliver fast, low-level, dexterous robot actions
  
  Everything in this video is a neural network:

Source, Figure AI LinkedIn post.

This week on my Robotics Newsletter I talked about Covariant, and Physical Intelligence seems to be hoping to do a very similar thing.

OK, Robot

An OpenAI Spinout is building a Robot Brain of AI

Hey Everyone, How do I say this? I think I started a Robotics Newsletter just in the nick of time. And how to put this? I’ll just say it bluntly: An OpenAI spinoff has built an AI model that helps robots learn tasks like humans. All of this just as I was thinking of the AI coding agent called…

Read more

2 years ago · 4 likes · 1 comment · Michael Spencer

Physical Intelligence Is Building a Brain for Robots

• Physical Intelligence's thesis is that the time is right for a new approach to building robotics AI models.

• The company looks to merge the techniques used to build language models with its own techniques for controlling and instructing machines.

Read More

So Figure AI, Covariant and Physical Intelligence are beginning to intersect the future of robotics with Generative AI.

While Figure AI appears to be the leader in physical robots, Covariant and Pi are connecting the dots. Figure announced that it had raised $675 million, which it said valued the company at $2.6 billion.

Towards a Universal Model for Smarter Robots

• Formed in 2024 itself by a team of robotics and AI experts, the company plans to create software that can add high-level intelligence to a wide variety of robots and machines.

• Or, as co-founder and Chief Executive Officer Karol Hausman  puts it in Physical Intelligence’s first public interview since founding the company: “We aim to bring AI to the physical world with a universal model that can power any robot or any physical device basically for any application.” 

Subscribe now

Read more

Hey Everyone,

There is another former Google AI startup startup that has recently caught my attention.

The DeepMind alums called their startup Haiper. Along the lines of OpenAI’s Sora, the company has developed an all-in-one visual foundation model for publishers, studios and individuals, enabling all to easily generate high-quality video content. 

Haiper, a London, UK-based video-generative AI platform building a foundation model for visual content, is coming out of stealth mode with $13.8M in seed funding.

Subscribe now

Haiper has been working on building an AI foundation model that can take a text prompt written by a user and transform it into a short, animated video. Users can also upload a static image alongside a written prompt and have it generate a creative animation.

Yet another team from Google DeepMind are using “AGI” in their mission statement.

Read more