🌐 REALTECH News, December 2024
+ Palantir on a tear, Trump 2.0, robotic dexterity and a robotaxi IPO
Hi all, what a crazy month! With Trump back in the White House, the global business and technology die is being recast. Trump’s nomination places some of the US tech sector’s most controversial characters at the wheel of US policy, with a focus on a more insular neo-liberal philosophy. US election certainty has triggered a defrosting of liquidity in venture markets. Numerous companies, such as ServiceTitan and Klarna, have filed ahead of imminent IPOs. More below
In today’s edition:
Trump 2.0, good for deep tech or deep trouble?
Palantir as a “Trump trade” and it’s bumper Q3 earnings
Meta releases research, pushing the future of robotics
The implementation challenges of AI in defence
📣 PSAs
🧑🏼🔬 Research: Alongside Dealroom, we have compiled 40+ pages of in-depth analysis on trends in REALTECH. We analysed 15k companies, 11k investors over 10 years - read it here
✈️ December travels, I will be in Zurich (Dec 5th - 7th) and Barcelona (Dec 11th) - HMU if you’re around!
If you’d like to listen to the audio version of this post, click here or subscribe via Spotify
Top Stories
Trump 2.0: no conflict, no interest
Trump is back. The day after Trump’s victory, the NASDAQ, a composite index of mostly tech stocks, soared by 2.9%. Trump is preparing to re-enter the White House, with the Republican Party likely controlling both the Senate and the House of Representatives. The dynamics this time are different. Trump is flanked by a new cadre of controversial right-leaning tech leaders, such as Elon Musk and Peter Thiel acolyte J.D. Vance, the VP in waiting. United by a neo-liberal, Javier Milei-esque philosophy of slashing government bureaucracy and eliminating perceived inhibitors to growth. Potentially ending globalisation, a hallmark of the post-WWII international trade order.
But philosophy is one thing; policy execution is another. Let’s look at how Trump’s philosophy might shape key sectors:
🟢 Regulation: unless you’ve been living under a rock, you would have heard of the new Department of Government Efficiency (DOGE), co-headed by Musk. DOGE has an 18-month mandate to cut regulations, the civil service, and government inefficiencies. This could unleash a wave of capital and entrepreneurial activity, particularly in financial services, crypto, and AI. It will also relieve the tech industry of Lina Khan’s anti-competitive oversight, paving the way for unfettered M&A. Trump has vowed to roll back Biden’s proposed AI regulation.
🟢 Defence: this is more complicated. The war in Ukraine has arguably been a boon for the US industrial base, gutting capacity from a beleaguered Germany. Trump views himself as much a peacemaker as a dealmaker, favouring negotiations with strongmen leaders and using his weight to pacify. To what extent he will engage with Putin and the situation in Ukraine remains unclear. He has made it clear that NATO countries must increase their contributions to the target 2% of GDP and suggested that European nations should support Ukraine largely without US involvement. This could end the post-Bretton Woods era, with America as the global protector, safeguarding security and facilitating international trade. Adding to the mixed picture. It’s rumoured a partner at Peter Thiel’s Founders Fund—known for investing in Anduril and Palantir—is rumoured to be replacing Kathleen Hicks as the Pentagon’s no.2. While still speculative, the rumour hints at the proximity of Trump’s administration to vested interests in defence.
🟠 Manufacturing: Trump’s first term delivered positively for US manufacturing through tax reform and renegotiated trade agreements. His second term is expected to double down on this agenda, with new incentives for reshoring critical industries through trade tariffs (more below). This is a long-term positive, though, in the short term, it may present volatility for manufacturers with complex international supply chains
🟠 Supply Chain: Trump is pushing to levy tariffs of 10-20% on imports, with levies as high as 60% on Chinese imports. If enacted this would be destructive to global growth. This is a huge issue for deeptech companies reliant on foreign and often cheap Chinese manufactured parts. Some companies have started buying imported goods in advance of Trump taking office in January. Though, what remains to be seen is whether this narrative is mere postering to enact leverage against other nation-states. A strategy that worked for him in the last presidency
🔴 Energy and climate: Trump is a climate sceptic, informed by his experience as a property developer battling environmental regulations. Trump has chosen Chris Wright, an oil and gas executive, to lead the Department of Energy. Wright is a proponent of oil and fracking and sits on the board of Sam Altman-backed nuclear company Oklo (NYSE:OKLO). His administration is expected to roll back the green tax credits from Biden’s Inflation Reduction Act (IRA) and dismantle federal ESG mandates. The iShares Global Clean Energy ETF is down 9 per cent since Trump won
While markets have popped after his nomination, markets rely on stability, which the first Trump presidency lacked. We’ll have to wait and see whether round two is any better. Notably, this wave of mass deregulation could serve as a bellwether for other developed nations, such as the EU, where famously burdensome regulatory frameworks have hindered innovation and competitiveness.
While the narrative and philosophy is to ‘enrich the forgotten,’ it seems likely that Trump proxies will be funnelling influence and capital to entrenched interests.
As the legendary venture capitalist, John Doerr said: ‘no conflict, no interest’
AI in defence
It’s been a busy month for AI’s role in national security. Firstly, data labelling company Scale AI announced ‘Defense Llama’, a forked version of Meta’s Llama 3 (LLM) trained on specific government data and the DoD’s Ethical Principles. Less than 24 hours later, Palantir and Anthropic announced their joint release of Anthropic’s Claude 3.5 LLM for national security professionals. Whilst AI is a critical new vector for defence departments and startups alike, its implementation is complex.
AI is having a moment, duh - A trend we’ve been covering is defence’s shift from siloed domains to multi-domain coordination amongst forces—going from ‘analog’ to ‘digital’. AI is taking a central, coordinative role in this new paradigm and being integrated across various capabilities. In agencies, operations systems, and on the battlefield, from command and control platforms to intelligence analysis tools and decision-support systems. The goal is better decisions and more capabilities, at machine speed.
The US, UK, and NATO have been vocal about adopting AI. The UK’s Ministry of Defence (MoD) published its AI Defence Strategy in 2022 under the Johnson government. The US DoD has been working on implementing AI since 2018, with a strategy refresh published last year. In 2022, the US DoD appointed its first Chief Digital and Artificial Intelligence Officer (CDAO) and initiated Project Linchpin. The latter is the DoD’s first Program of Record to focus on AI. Delivering a full AI pipeline for the Army (as per below) that can integrate with private sector software and models from trusted vendors such as Palantir and Scale AI. Easy right?
If not now, then when? Implementing AI is more urgent now than ever as geopolitical tensions mount and the gap between China and the West’s computing capabilities narrows. However, there are many issues with AI adoption - such as a lack of digital infrastructure, perverse incentives, outdated procurement paradigms and test-and-evaluation (T&E) processes:
“That sort of continuous capability development in a software environment is a real challenge culturally to the department because our processes are gates for hardware programs,” Lt. General Michael Groen, Joint Artificial Intelligence Center (JAIC) Director
Most notably, setting AI systems’ ethical, safety and regulatory parameters remains complex. Ethics and safety are needed to define the lethality and responsibility of AI systems. The UK’s MoD just released the ‘Dependable AI in Defence’ directive, which builds upon the prior Defence AI Strategy:
‘[the directive] provides clear direction on how to implement the MOD’s AI ethical principles, to deliver safe, robust and effective AI-enabled capabilities which have the right level of human oversight to advance our military edge.’
Old systems getting an AI glow-up—Initial integration has been focused on improving organisational efficiencies and upgrading existing vehicles and platforms. Such as Palantir’s MAVEN data fusion system and Helsing’s AI enablement of the Eurofighter and GripenE fighter jets.
"Rather than identify a handful of AI-enabled warfighting capabilities that will beat our adversaries, our strategy outlines the approach to strengthening the organizational environment within which our people can continuously deploy data analytics and AI capabilities for enduring decision advantage," DOD Chief Digital and AI Officer (CDAO) Craig Martell
This adoption pattern allows forces to integrate, understand, and assess AI systems in a familiar form factor. However, these largely ‘human in the loop’ systems are a stepping stone to the end game.
Autonomy at the edge. Autonomy is critical not only due to drone-enabled asymmetric warfare (e.g., a $5000 Ukrainian drone destroying a $20m Russian tank) but also because battlefields are increasingly EW-jammed / GPS-denied or ‘Disconnected, Degraded, and Limited Bandwidth’ (DDIL) environments in military parlance. While it's arguably the end goal, operating swarms of autonomous robots that can kill in a disconnected environment is tricky, as you can imagine. Currently, military technologists are approaching it with caution.
Palantir on a tear: Trump, Q3 earnings, AI
“A juggernaut is emerging. This is the software century, and we intend to take the entire market. The unrelenting march of our business has been driven by an early and decades-long investment in the technical infrastructure that is now making LLMs that have reshaped our world useful and valuable to large enterprises.” Alex Karp, the Palantir CEO’s Q3 earnings letter to investors
Palantir finds itself in a unique position - at the intersection of the Trump presidency, AI and software adoption in critical industries.
Through 20 years of graft, Palantir has focused on the painful integration of software into the defense industrial base, and the development of their AIP and Foundry products. They are now reaping those rewards. Its stock ripped after their Q3 earnings - they announced sequential quarterly topline revenue growth acceleration as well as improving profitability (chart below):
![Q3 2024 Earnings Letter - Graph 1 [En]](https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb9864a9d-dbcb-4bab-8d51-5d94f1abe803_1800x877.png "Q3 2024 Earnings Letter - Graph 1 [En]")
Palantir proclaims itself as the first software defense prime, and with good reason. They have delivered AI systems such as TITAN ($178m - autonomous target recognition), MAVEN ($480m - sensor fusion system), and just this month, their drone visual navigation technology (VNav).
Today, Palantir’s biggest customer is the US government, providing software services to the NSA, CIA and numerous DoD operations. With Trump and J.D Vance in, investors are betting that the flow of government money will grow larger. Would you bet against that?
Meta drops research on the future of robotics
Meta just released a bunch of open-source robotic dexterity and manipulation research. The work focuses on allowing robots to ‘feel’, giving robotic end-affectors tactile sensing capabilities, making robots more human.
Notably, this points to the arc of progress for robotic systems from the past, present and future, from doing → thinking → sensing.
Touch is a crucial sensing modality that provides rich information about object properties and interactions with the physical environment. Something inherent to humans. This is particularly important in use cases that require delicate handling of properties - such as fruit picking, material handling and surgical robotics. Meta dropped four pieces of work:
Sparsh introduces a novel vision-based approach to tactile sensing using self-supervised learning to pre-train models on over 460,000 tactile images. This enables superior generalisation across various sensors and tasks while reducing reliance on labelled data
Digit 360 is an artificial finger-shaped tactile sensor with more than 18 sensing features. The sensor has over 8 million taxels for capturing omnidirectional and granular deformations and forces on the fingertip surface.
Digit Plexus aims to create a standardized platform that provides a hardware-software solution to integrate tactile sensors on any robot hand
PARTNR is a collaboration benchmark designed to evaluate how effectively AI models assist humans with household tasks. Built on top of Meta’s Habitat simulation platform, it features 100,000 natural language tasks across 60 virtual homes and incorporates over 5,800 unique objects. The benchmark measures the capability of LLMs and VLMs to understand and execute human instructions effectively
To unlock greater applications of embodied AI - we need robotic dexterity and sensing. However, the challenge is real. There is increasing work in this space. with various public and private research attempts to unlock further capabilities. Notably, open source work such as Stanford and Deepmind’s ALOHA (A Low-cost Open-source Hardware System for Bimanual Teleoperation) and Carnegie Mellon’s 3D printed DASH hands (dexterous anthropomorphic soft hand). The UK’s Advanced Research and Invention Agency (ARIA) has leaned heavily into this field. Recently launching the Robotic Dexterity Programme, run by Jenny Read. We’ve also seen development from well-funded humanoid companies such as Figure and 1x, as well as more end-effector-focused companies such as ETH Zurich spin-out Mimic.
🤓 Other stories you need to know
🟠 Y Combinator published their ‘Request for Startups’, which thematically is heavily leaning towards physical industries (Y Combinator)
🏭 IT/OT convergence: The 27 themes that define the future of industrial integration (IoT Analytics)
🛠️ ServiceTitan, the SaaS for tradesmen, is going public; this is the best S1 breakdown (Meritech Capital)
🌎 PokemonGo maker Niantic, has used geospatial data from gamers to build a Large Geospatial Model (LGM) to help devices understand and navigate the real world (Niantic)
🟢 Everything that happened at COP29 (Carbon Brief)
🚖 Chinese autonomous vehicle company, PonyAI, went public at $5bn market cap (UK investor magazine)
🍟 The Biden administration fast-tracked a $11bn payment to TSMC, ahead of Trump’s arrival in the White House (Financial Times)
🚁 Skydio lost a key government contract to drone maker Red Cat, and investors stepped in to sure up its balance sheet (DroneXL)
💰Notable Funding Rounds
Odoo ($527M secondaries) offers open-source ERP solutions for businesses, lifting its valuation to $5.26B. CapitalG and Sequoia led the funding
Skydio ($170M Series E extension) manufactures autonomous drones for commercial and defense use, backed by Linse Capital
Chaos Industries ($145M Series B) develops advanced defense technologies for military applications, with funding led by Accel
Radiant Nuclear ($100M Series C) is developing portable nuclear microreactors for clean energy solutions, led by DCVC
Tekever ($74M) builds dual-use drone platforms for defense and commercial applications, with funding led by Bailie Gifford and NATO Innovation Fund
Cradle ($73M Series B) uses AI for protein design to accelerate biotech innovation, led by Insight Partners
Roboflow ($40M Series B) provides AI tools for computer vision development, with funding led by GV
Neara ($31M Series C) provides digital twin solutions for infrastructure projects, led by EQT
Wherebots (€21M Series A) builds AI solutions for large-scale industrial and environmental optimization. The round was led by Felicis
BrightAI ($15m seed) designs physical AI solutions and reached $80M in annual revenue without external funding, Upfront Ventures led the seed
Flipturn ($11M Series A) focuses on EV fleet charging and management solutions, backed by CRV
Keel ($6M Seed) creates next-generation operational software for maritime industries, led by local globe and Earlybird
Emidat (€4M seed) simplifies emissions management in construction projects, backed by General Catalyst
Algorized ($4.3M in Seed) focuses on AI-driven optimization tools for supply chain management, with funding led by Amazon
Molyon (€4.3M Seed) advances synthetic biology for biotech and manufacturing, led by Plural and IQ Capital
HyperHeat (€3.5M seed) develops industrial heating solutions to decarbonize processes, led by Amadeus APEX Technology Fund