![\"Green](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2025\/06\/1751219116_600_960x0.jpg\")
<\/p>\n<\/p>\n
microchip integrated on motherboard<\/p>\n
getty<\/p>\n
In the game of AI acceleration, there are several key moving parts. One of them is hardware: what do the chips look like? And this is a very interesting question. Another is quantum computing: what role will it play? Another is scaling.<\/p>\n
Everyone from CEOs and investors to engineers is scrambling to figure out what the future looks like, but we got a few ideas from a recent panel at Imagination in Action that assembled some of the best minds on the matter.<\/p>\n
WSE and the Dinner Plate of Reasoning<\/p>\n
Not too long ago, I wrote about the Cerebras WSE chip, a mammoth piece of silicon about the size of a dinner plate, that is allowing the centralization of large language model efforts. This is an impressive piece of hardware by any standard, and has a role in coalescing the vanguard of what we are doing with AI hardware.<\/p>\n
In the aforementioned panel discussion, Julie Choi from Cerebras started by showing off the company\u2019s WSE superchip, noting that some call it the \u201ccaviar of inference.\u201d (I thought that was funny.)<\/p>\n
\u201cI think that as we evolve, we’re just going to see even more innovative, novel approaches at the hardware architecture level,\u201d she said.<\/p>\n
\u201cThe optimization space is extremely large,\u201d said Dinesh Maheshwari, discussing architecture and compute units. \u201cSo I encourage everyone to look at it.\u201d<\/p>\n
Panelist Caleb Sirak, also of MIT, talked about ownership of hardware.<\/p>\n
\u201cAs the models themselves start to change, how can businesses themselves integrate them directly and get them for a fair price, but also convert that AI, and the energy involved, into a productive utility?\u201d<\/p>\n
\u201cWhat is a computer, and what can a computer do?\u201d asked Alexander Keesling, explaining his company\u2019s work on hardware. \u201cWe took the fundamental unit of matter, a single atom, and turned it into the fundamental unit of information, which is a quantum bit \u2026 a quantum computer is the first time in human history where we can take advantage of the fundamental properties of nature to do something that is different and more powerful.\u201d<\/p>\n
Jeremy Kepner of MIT\u2019s Lincoln Lab had some thoughts on the singularity of computing \u2013 not the race toward AGI, but a myopic centralization of an overarching \u201coperation.\u201d<\/p>\n
\u201cEvery single computer in the high end that we built for the last many decades has only done one operation,\u201d he said. \u201cSo there’s a lot to unpack there, but it’s for very deep mathematical and physics reasons: that’s the only operation we’ve ever been able to figure out how to accelerate over many decades. And so what I often tell the users is, the computer picks the application. AI happens to be acceleratable by that operation.\u201d<\/p>\n
He urged the audience to move forward in a particular way.<\/p>\n
\u201cThink about whatever you want to do, and if you can accelerate it with that kind of mathematical operation, you know the sky is the limit on what you can do,\u201d he said. \u201cAnd someone in your field will figure it out, and they will move ahead dramatically.\u201d<\/p>\n
Engineering Challenges and AI Opportunities<\/p>\n
The panel also mentioned some of the headwinds that innovators must contend with.<\/p>\n
On the other hand, Jeff Grover noted the near-term ability of systems to evolve.<\/p>\n
\u201cWe’re actually quite excited about this,\u201d he said.<\/p>\n
The Software End<\/p>\n
Panelists discussed the relevance of software and the directions that coding is going in.<\/p>\n
\u201cProgramming languages are built for people,\u201d Sirak said. \u201cHow do you actually change that to build languages and tools that AI can use?\u201d<\/p>\n
Choi mentioned benchmarks like inference rates of 2900 tokens per second for Llama 4.<\/p>\n
\u201cOpen source models are rich for developers,\u201d she said. \u201cWhat that’s doing is building a bridge between the bravest developers. I would say the early adopters tend to be very courageous, and they’re willing to code on things that they’ve never seen before.\u201d<\/p>\n
The Fast Car<\/p>\n
Several panelists talked about a particular metaphor to a Ferrari, with Choi referencing \u201cFerrari-level\u201d speeds for the Cerebras chip.<\/p>\n
Maheshwari talked about \u201cexotic\u201d chips, and design from an architecture paradigm, comparing certain builds to \u201cpicking up groceries in a Ferrari.\u201d<\/p>\n
He also mentioned the imperative of keeping the technology \u201cstreet legal.\u201d<\/p>\n
Moore\u2019s Law and Progress<\/p>\n
Kepner talked about being surprised by what computers can do, and the size of investment in the industry. Moore\u2019s law, he said, implied an upper limit for spending. He predicted another decade of efficiencies, and cited the Ozaki scheme, a matrix method for preserving precision in calculations.<\/p>\n
What About Quantum?<\/p>\n
\u201cI think that the first area where we’re going to see quantum computing impact is going to be in research,\u201d Keesling said. \u201cThese problems, at their core, are (about) trying to answer what happens when atoms and electrons interact with one another and develop these emergent behaviors \u2026 how we think about chemistry, how we think about drug interactions, how we think about material properties, all comes from electrons and atoms moving.\u201d<\/p>\n
There was a lot to unpack in this panel discussion, including details on how we\u2019re going to achieve progress in the next few years.<\/p>\n
The Ozaki Scheme<\/p>\n
Going back to this matrix idea, I was not familiar with this term, so I looked it up and asked ChatGPT to describe it in basic English.<\/p>\n
\u201cIt\u2019s named after Makoto Ozaki, the person who came up with the idea,\u201d the model told me. \u201cHe found a smart way to do very accurate math (like multiplying big grids of numbers) using fast but less accurate tools (like low-precision numbers). His method splits the work into small, simple steps and then carefully puts the pieces back together to get the exact right answer.\u201d<\/p>\n
Going further, ChatGPT, just to be nice, even gave me a medieval storyline to show how the Ozaki scheme works, and to contrast it to other alternatives.<\/p>\n
I\u2019m just going to print that here, because it\u2019s interesting.<\/p>\n
The Tale of the Kingdom of Matrixland<\/p>\n
In the kingdom of Matrixland, the royal court has a big job: multiplying giant tables of numbers (called matrices). But the royal calculator is slow when it uses fancy, high-precision numbers.<\/p>\n
So the King holds a contest:
\u201cWho can multiply big matrices both quickly and accurately?\u201d<\/p>\n
Sir Ozaki\u2019s Clever Trick<\/p>\n
Sir Ozaki, a wise mathematician, enters the contest.
He says:<\/p>\n
\u201cI\u2019ll break each matrix into small, easy pieces that the royal calculator can handle quickly. Then I\u2019ll multiply those simple parts and put them back together perfectly.\u201d<\/p>\n
The crowd gasps!
His method is fast and still gives the exact right answer.<\/p>\n
The King declares it the Ozaki Scheme.<\/p>\n
The Other Contestants<\/p>\n
But other knights have tricks too:<\/p>\n
Lady Refina (Iterative Refinement)
She does the quick math first, then checks her work.
If it\u2019s off, she fixes it \u2014 again and again \u2014 until it\u2019s just right.
She\u2019s very accurate, but takes more time.<\/p>\n
Sir Compenso (Compensated Summation)
He notices small errors that get dropped during math and catches them before they vanish.
He\u2019s good at adding accurately, but can\u2019t handle full matrix multiplication like Ozaki.<\/p>\n
Lady Mixie (Mixed Precision)
She charges in with super speed, using tiny fast numbers (like FP8 or FP16).
Her answers aren\u2019t perfect, but they\u2019re \u201cgood enough\u201d for training the kingdom\u2019s magical beasts (AI models).<\/p>\n
Baron TensorFloat (TF32)
He uses a special number format invented by the kingdom\u2019s engineers.
Faster than full precision, but not as sharp as Ozaki.
A favorite of the castle\u2019s GPU-powered wizard lab.<\/p>\n
The Ending<\/p>\n
Sir Ozaki\u2019s method is the most exact while still using fast tools.
Others are faster or simpler, but not always perfect.<\/p>\n
The King declares:<\/p>\n
\u201cAll of these knights are useful, depending on the task. But if you want both speed and the exact answer, follow Sir Ozaki\u2019s path!\u201d<\/p>\n
Anyway, you have a range of ideas here about quantum computing, information precision, and acceleration in the years to come. Let me know what you think about what all of these experts have said about the future of AI.<\/p>\n","protected":false},"excerpt":{"rendered":"microchip integrated on motherboard getty In the game of AI acceleration, there are several key moving parts. One…\n","protected":false},"author":2,"featured_media":224560,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3164],"tags":[3284,53157,53,16,15],"class_list":{"0":"post-224559","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-computing","8":"tag-computing","9":"tag-enterprise-tech","10":"tag-technology","11":"tag-uk","12":"tag-united-kingdom"},"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@uk\/114767903405249263","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts\/224559","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/comments?post=224559"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts\/224559\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/media\/224560"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/media?parent=224559"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/categories?post=224559"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/tags?post=224559"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}